INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Activation of G-protein-coupled, group I metabotropic glutamate receptors (mGluRs) by either synaptically released glutamate (Bolshakov and Siegelbaum 1994; Kemp and Bashir 1999; Oliet et al. 1997; Otani and Connor 1998) or by pharmacological agonists such as (R,S)-3,5-dihydroxyphenylglycine (DHPG) (Fitzjohn et al. 1999; Huber et al. 2000, 2001; Kleppisch et al. 2001; Palmer et al. 1997) elicits a long-term depression (LTD) of synaptic transmission at excitatory synapses onto hippocampal CA1 pyramidal cells. DHPG also induces LTD at medial perforant path synapses onto granule cells in the dentate gyrus (Comodeca et al. 1999). While recent reports have shown that DHPG-induced LTD in the hippocampal CA1 region is dependent on activation of G_q-type G proteins (Kleppisch et al. 2001) and protein synthesis (Huber et al. 2000), relatively little is known about the mechanisms responsible for DHPG-induced LTD in the hippocampal CA1 region (however, see Schnabel et al. 1999a,b, 2001). Indeed, it is not yet clear whether the synaptic depression induced by DHPG in hippocampal CA1 pyramidal cells arises from pre- and/or postsynaptic mechanisms. Several findings, including the postsynaptic localization of group I mGluRs at these synapses (Luján et al. 1996; Romano et al. 1995; Shigemoto et al. 1997) and the fact that postsynaptic injection of protein synthesis inhibitors into CA1 pyramidal cells blocks DHPG-induced LTD (Huber et al. 2000), indicate that postsynaptic mechanisms are primarily involved. Other findings suggest, however, that activation of group I mGluRs with DHPG depresses transmission through a presynaptic inhibition of transmitter release (Gereau and Conn 1995; Herrero et al. 1998; Mannaioni et al. 2001; Manzoni and Bockaert 1995; Rodriguez-Moreno et al. 1998). Finally, DHPG-induced LTD might depend on both pre- and postsynaptic processes because studies using hippocampal slices from young rats suggest that a combination of postsynaptic induction and presynaptic expression is involved in mGluR-dependent forms of LTD induced by synaptic stimulation (Bolshakov and Siegelbaum 1994; Oliet et al. 1997). Here we investigated the synaptic locus of DHPG-induced LTD in the hippocampal CA1 region by using a combination of postsynaptic injections of pharmacological reagents to examine the role of postsynaptic processes along with extracellular manipulations designed to probe the potential involvement of presynaptic mechanisms. Our results suggest that activation of postsynaptic mGluRs with DHPG depresses excitatory synaptic transmission through changes in presynaptic Ca²⁺ signaling and thus indicate that retrograde signaling via an as yet unidentified messenger(s) has an important role in this form of LTD.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Standard techniques were used to prepare 400-µm-thick hippocampal slices from tissue obtained from 5- to 7-wk-old male C57BL/6 mice that had been anesthetized with halothane prior to being killed by cervical dislocation. Slices were maintained at 30°C in an interface-type chamber (Fine Science Tools, Foster City, CA) that was perfused with an oxygenated (95% O₂-5% CO₂) mouse artificial cerebrospinal fluid (ACSF) consisting of (in mM) 124 NaCl, 4.4 KCl, 25 NaHCO₃, 1.0 NaH₂PO₄, 1.2 MgSO₄, 2.0 CaCl₂, and 10 glucose (flow rate = 2-3 ml/min). Slices were allowed to recover for 2 h prior to an experiment. For an experiment, a slice was placed in a submerged-slice recording chamber and a bipolar stimulating electrode was used to activate Schaffer collateral/commissural (SC) fiber synapses onto CA1 pyramidal cells and associational/commissural fiber synapses (AC) onto CA3 pyramidal cells. The resulting field excitatory postsynaptic potentials (fEPSPs) were recorded using low-resistance glass microelectrodes (5-10 M, filled with ACSF) placed in stratum radiatum of either the CA1 or CA3 region. Presynaptic fiber stimulation pulses were delivered at 0.02 Hz using a stimulation intensity that evoked fEPSPs that were 50% of the maximal fEPSP amplitude.

Whole cell current-clamp recordings were used to record EPSPs from individual CA1 pyramidal cells and to introduce reagents selectively into postsynaptic cells. Low-resistance electrodes (2-5 M, access resistance ranged from 9 to 28 M) were filled with a solution containing (in mM) 127.5 K-gluconate, 17.5 KCl, 1.0 MgCl₂, 0.2 EGTA, 10 HEPES, 2 Mg-ATP, and 0.3 GTP (pH = 7.2). In some experiments, we used an electrode-filling solution designed to block postsynaptic K⁺ channels that contained (in mM) 122.5 Cs-gluconate, 17.5 CsCl, 10 TEA-Cl, 0.2 EGTA, 10 HEPES, 2 Mg-ATP, and 0.3 GTP (pH = 7.2). If needed, cells were hyperpolarized to between 70 and 75 mV using constant current injection through the recording electrode. Throughout the experiment a 50-ms-long pulse of hyperpolarizing current (0.1 nA) was injected 150 ms after each EPSP to monitor input and access resistance. Presynaptic fiber stimulation pulses were delivered once every 20 s using a stimulation intensity sufficient to evoke EPSPs between 10 and 15 mV in amplitude. Whole cell recordings were maintained for 18-20 min before DHPG application to allow diffusion of substances in the electrode solution into the cell.

Whole cell voltage-clamp recordings were used to record excitatory postsynaptic currents (EPSCs) in CA1 pyramidal cells. In these experiments, the slices were bathed in a modified ACSF containing 100 µM picrotoxin to block GABA_A receptor-mediated inhibitory postsynaptic currents, and the concentration of KCl was lowered to 2.4 mM. In addition, the CA3 region of the slice was removed to prevent bursting. Patch-clamp electrodes were filled with the Cs-gluconate electrode-filling solution described in the preceding text, and cells were voltage-clamped at 60 mV. Presynaptic stimulation pulses were delivered once every 20 s and a 30-ms hyperpolarizing voltage step (2 mV) was delivered 50 ms before each pulse of synaptic stimulation to monitor input and access resistance throughout the experiment. 6-cyano-7-nitroquionoxaline-2,3-dione (CNQX, 10 µM) was added to the ACSF to block the AMPA receptor-mediated component of EPSCs in experiments where the effects of DHPG on NMDA receptor-mediated EPSCs were investigated.

Data acquisition and analysis were performed using Experimenter's Workbench and Common Processing software package (Data Wave Technologies, Longmont, CO). All values are reported as means ± SE. Unless indicated otherwise, a 10-min bath application of 100 µM DHPG was used in all experiments to reliably induce a robust and persistent depression of synaptic transmission. The effects of DHPG on synaptic transmission were assessed using the average size of synaptic responses recorded over the last 5 min of a 10-min application to determine the initial effects of DHPG on synaptic transmission. The average size of synaptic responses recorded between 25 and 30 min after DHPG washout was used to determine the persistent effects of DHPG on synaptic transmission. Paired t-tests and one-way repeated-measure ANOVAs (followed by Dunnett's test comparisons) were used to determine statistical significance for within-group comparisons. Unpaired t-tests or, where appropriate, one-way ANOVAs (followed by Dunnett's tests) were used for between-group comparisons.

DHPG, (1S,3R)-1-aminocyclopentane-1,3-dicarboxylic acid (ACPD), and 2S-2-amino-2-(1S,2S-2-carboxycyclopropyl-1-yl)-3(xanth-9-yl) propanoic acid (LY341495) were obtained from Tocris Cookson (Ballwin, MO). All other compounds were from Sigma (St. Louis, MO).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As reported previously (Fitzjohn et al. 1999; Palmer et al. 1997), activation of mGluRs with DHPG induced a strong depression of synaptic transmission at Schaffer collateral/commissural fiber synapses in area CA1 that persisted for 30 min following DHPG washout (Fig. 1A). In contrast, the mGluR agonist ACPD induced only a transient suppression of synaptic transmission that fully recovered when ACPD was washed from the recording chamber with agonist-free ACSF (Fig. 1B). DHPG, but not ACPD, also induced LTD at associational/commissural fibers synapses onto pyramidal cells in the CA3 region (Fig. 1, C and D). While the effects of DHPG on synaptic transmission in both the CA1 and CA3 regions were strongly inhibited by a high concentration of the mGluR antagonist LY341495 (Figs. 1, A and C, and 2A), the N-methyl-D-aspartate (NMDA) receptor antagonist D,L-2-amino-5-phosphonovaleric acid (APV, 100 µM) had no effect on DHPG-induced LTD in the CA1 region (n = 5, data not shown). Thus similar to previous observations in rat hippocampus (Huber et al. 2001; but see Palmer et al. 1997), activation of mGluRs with DHPG induces an NMDA receptor-independent form of LTD in mouse hippocampal slices.

View larger version (35K): [in this window] [in a new window]   Fig. 1. Activation of group I metabotropic glutamate receptors (mGluRs) with (R,S)-3,5-dihydroxyphenylglycine (DHPG) but not ACPD induces a lasting depression of excitatory synaptic transmission in both the CA1 and CA3 regions of the hippocampus. A: following a 20-min period of baseline recording, 100 µM DHPG was bath applied for 10 min (; DHPG application indicated by ). DHPG induced a strong depression of synaptic transmission [field excitatory postsynaptic potentials (fEPSPs) were depressed to 28.3 ± 6.8% of baseline during the last 5 min of the DHPG application, n = 5] that showed only partial recovery following DHPG washout (30 min after DHPG washout fEPSPs were depressed to 53.4 ± 6.7% of baseline). Both the initial and persistent depression induced by DHPG were blocked in experiments where the mGluR antagonist 2S-2-amino-2-(1S,2S-2-carboxycyclopropyl-1-yl)-3(xanth-9-yl) propanoic acid (LY341495, 20 µM) was applied for 20 min beginning 5 min before DHPG application (, n = 5). Experiments were done in the CA1 region of the hippocampus. Inset: example fEPSPs recorded during baseline (larger response) and 30 min after DHPG washout (smaller response). Calibration bars are 0.5 mV and 5 ms. B: in the CA1 region a 10-min bath application of 100 µM ACPD induced a robust depression of synaptic transmission that fully recovered following washout of ACPD from the recording chamber (n = 5, ACPD application indicated by the ). C: a 10-min bath application of 100 µM DHPG induced long-term depression (LTD) of association/commissural (AC) fiber synapses in the hippocampal CA3 region (, n = 5). During the last 5 min of DHPG application, fEPSPs were reduced to 18.4 ± 4.5% of baseline and were 26.9 ± 5.5% of baseline 30 min after DHPG washout. , experiments (n = 4) where DHPG was applied in the presence of LY341495 (20 µM, LY341495 was present for 20 min beginning 5 min before application of DHPG). Inset: example fEPSPs recorded during baseline (larger response) and 30 min after DHPG washout (smaller response). Calibration bars are 0.5 mV and 5 ms. D: as in the CA1 region, a 10-min application of 100 µM ACPD induced a robust, but reversible, depression of AC fiber synapses in the CA3 region (n = 5).

View larger version (31K): [in this window] [in a new window]   Fig. 2. The mGluR antagonist LY341495 blocks the induction and expression but not the maintenance of DHPG-induced LTD. A: results from a single experiment in the CA3 region showing the block of DHPG-induced LTD by LY311495. , application of DHPG (100 µM); , the presence of LY341495 (20 µM) in the recording chamber. Inset: fEPSPs recorded at the times indicated by the numbers. Calibration bars are 0.5 mV and 5 ms. B: results from a single experiment in the CA3 region showing that LY341495 has no effect on basal synaptic transmission at AC fiber synapses but completely reverses DHPG-induced LTD (, LY341495 application; , DHPG application). Note that the depression induced by DHPG is re-established following washout of LY341495. Inset: fEPSPs recorded at the indicated times. Calibration bars are 0.5 mV and 5 ms. C and D: blockade of mGluRs with LY341495 inhibits the expression but not the maintenance of DHPG-induced LTD in both the CA3 (C) and CA1 (D) regions of the hippocampus. Average results from 5 experiments in the CA3 region and 5 experiments in the CA1 region where LY341495 was applied for 10 min beginning 20 min after the induction of LTD by DHPG ( and , presence of LY341495 and DHPG, respectively, in the bath).

The ability of LY341495 to inhibit both the transient and persistent depression of synaptic transmission induced by DHPG indicates that activation of mGluRs is required for the induction of this form of LTD. To examine whether mGluR activation is also required for the expression and/or maintenance of DHPG-induced LTD, we investigated whether blocking mGluRs with the LY341495 had any effect on the depression of synaptic transmission induced by prior DHPG application. In both the CA3 and CA1 regions, the depression of synaptic transmission induced by DHPG was strongly reversed when LY341495 was applied for 10 min beginning 20 min after DHPG was washed out of the recording chamber (Fig. 2, B-D). This result, which is similar to previous findings in rat hippocampal slices (Fitzjohn et al. 1998; Palmer et al. 1997), indicates that activation of mGluRs is required for the expression of DHPG-induced LTD. Following washout of the LY341495, however, the synaptic depression was re-established nearly intact (Fig. 2). Thus persistent activation of mGluRs does not seem to be required for the maintenance of DHPG-induced LTD because LTD reappears after mGluR antagonists are washed from the recording chamber (see also Fitzjohn et al. 1998; Palmer et al. 1997).

To determine whether activation of postsynaptic mGluRs is specifically required for both the initial and persistent inhibition of synaptic transmission induced by DHPG, we performed whole cell recordings from CA1 pyramidal cells using a potassium gluconate-based electrode-filling solution where GTP was replaced with 1 mM guanosine-5'-O-(2-thiodiphosphate) (GDPS) to inhibit activation of postsynaptic G proteins. While DHPG induced a robust and persistent depression of synaptic transmission in interleaved control recordings with GTP containing solutions, only a small initial depression and no persistent depression was observed in recordings where GDPS was present in the electrode-filling solution (Fig. 3A). To control for the possibility that loading postsynaptic cells with GDPS by itself might have effects on synaptic transmission, we monitored evoked EPSPs for 60 min starting within 1 min after obtaining whole cell recordings with GDPS-containing electrodes. In all cells studied (n = 5), EPSPs grew dramatically (4-fold on average) during the first 5-10 min of whole cell recording with GDPS-containing electrodes perhaps due to effects of GDPS on AMPA receptor endocytosis (Lüscher et al. 1999) and/or clearance from the slice of substances that leaked out of the electrode before seal formation. Following this initial "run-up," however, synaptic transmission was stable for the remainder of the experiment. When normalized to a baseline determined by the average size of EPSPs recorded between 10 and 20 min of whole cell recording (which corresponds to the baseline used in the experiments shown in Fig. 3A), EPSPs recorded after 25-30 min of whole cell recording were 98.5 ± 3.3% of baseline, whereas those recorded after 55-60 min of whole cell recording were 107.4 ± 4.3% of baseline. This indicates that loading postsynaptic cells with GDPS does not induce a slow onset enhancement of synaptic transmission that might mask a DHPG-induced depression. As an additional control, we also examined the effects of postsynaptic GDPS on the inhibition of synaptic transmission by adenosine, which is known to regulate synaptic transmission at excitatory synapses in the CA1 region through presynaptic effects on transmitter release (Lupica et al. 1992; Prince and Stevens 1992; Scholz and Miller 1991; Wu and Saggau 1994). While postsynaptic GDPS completely prevented the postsynaptic hyperpolarization and decrease in input resistance induced by adenosine (data not shown), it had no effect on the depression of synaptic transmission induced by a 10-min bath application of 100 µM adenosine (Fig. 3B). Together, these results suggest that both the initial and the persistent depression induced by DHPG are dependent on postsynaptic, GTP-dependent signaling pathways.

View larger version (20K): [in this window] [in a new window]   Fig. 3. Blocking postsynaptic G proteins with guanosine-5'-O-(2-thiodiphosphate) (GDPS) suppresses both the initial and persistent synaptic depression induced by DHPG. A: whole cell current-clamp recordings were used to record EPSPs from individual CA1 pyramidal cells using electrode-filling solutions that contained either GTP (, n = 12) or 1.0 mM GDPS (, n = 6). In cells where the electrode-filling solution contained GTP, EPSPs were depressed to 47.1 ± 3.2% of baseline in the presence of DHPG (indicated by ) and were still depressed to 57.4 ± 5.2% of baseline 30 min after DHPG washout. DHPG had little effect on synaptic transmission in interleaved experiments where the electrode-filling solution contained GDPS (EPSPs were 91.4 ± 4.3% of baseline in the presence of DHPG and 112.9 ± 2.6% of baseline following DHPG washout). Inset: example EPSPs (average of 3 responses) recorded during baseline and 30 min after DHPG washout in a control experiment (left) and in an experiment where GDPS was present in the electrode (right). Calibration bars are 2.5 mV and 15 ms. B: postsynaptic GDPS does not block the inhibition of excitatory synaptic transmission induced by bath application of adenosine. EPSPs were depressed 27 ± 6.5% of baseline during the last 5 min of a 10-min application of 100 µM adenosine in control cells (, n = 7) and were depressed to 20.4 ± 5% of baseline in cells where the patch-clamp electrode was filled with a solution containing 1.0 mM GDPS (, n = 8). Inset: EPSPs (average of 3 responses) recorded before and at the end of the adenosine application in control experiments (left) and in experiments where GDPS was present in the electrode solution (right). Calibration bars are 2.5 mV and 10 ms.

Previous studies suggest that activation of group I mGluRs with DHPG depresses synaptic transmission through presynaptic effects on transmitter release, perhaps due to an inhibition of voltage-sensitive Ca²⁺ channels (Gereau and Conn 1995; Herrero et al. 1998; Rodriguez-Moreno et al. 1998). Thus although our experiments with postsynaptic injection of GDPS indicated that postsynaptic processes are responsible for the synaptic depression induced by DHPG, we also investigated whether changes in presynaptic Ca²⁺ signaling might be involved. In these experiments, we used the approach outlined by Wheeler et al. (1996) in their study of the role of specific subtypes of voltage-dependent Ca²⁺ channels in synaptic transmission at Schaffer collateral/commissural fiber synapses onto CA1 pyramidal cells. As described by Wheeler et al. (1996), the instantaneous Ca²⁺ flux into the presynaptic terminals (F) can be described by the equation

(1)

where N_tot is the total number of Ca²⁺ channels, P_o is the probability of channel opening, and i_Ca²⁺ is the Ca²⁺ flux through a single channel. While we have not used this relationship in a quantitative manner to test for potential presynaptic effects of DHPG on synaptic transmission, the fact that the relationship in Eq. 1 is multiplicative indicates that a constant value of F can be maintained if changes in one parameter are compensated for by changes in the other parameters. Thus Eq. 1 makes several qualitative predictions regarding how the DHPG-induced depression of synaptic transmission might be affected by experimental manipulations of N_tot, P_o, and/or i_Ca²⁺ if DHPG depresses synaptic transmission through effects on presynaptic calcium signaling. For instance, if DHPG acts by decreasing P_o and/or by reducing the total number of channels available for activation (N_tot), Eq. 1 predicts that increasing i_Ca²⁺ by increasing extracellular Ca²⁺ concentrations ([Ca²⁺]_o) should be able to attenuate the DHPG-induced depression of synaptic transmission. Conversely, under conditions of low-[Ca²⁺]_o (where i_Ca²⁺ is reduced), decreases in P_o or N_tot induced by DHPG should have an even more dramatic effect on synaptic transmission as now even small changes in either parameter could be sufficient to reduce Ca²⁺ influx to levels below that needed to support transmitter release.

Because adenosine depresses synaptic transmission by inhibiting presynaptic calcium channels (Wu and Saggau 1994, 1997), we first examined the effects of changes in i_Ca²⁺ on the depression of synaptic transmission induced by adenosine to determine whether this approach might be useful for examining potential presynaptic changes in DHPG-induced LTD. In these experiments, we examined the effects of adenosine (25 µM) on synaptic transmission in slices bathed in a modified ACSF where the concentration of CaCl₂ was either decreased to 1.0 mM (low-Ca²⁺ ACSF) or increased to 6.0 mM (high-Ca²⁺ ACSF). As expected from previous work showing that adenosine inhibits transmitter release at these synapses through effects on presynaptic calcium channels and as predicted by Eq. 1, the depression induced by adenosine was significantly suppressed in slices bathed in high-Ca²⁺ ACSF and significantly enhanced in slices bathed in a low-Ca²⁺ ACSF (Fig. 4A). Changes in extracellular Ca²⁺ levels had a similar and even more striking affect on the ability of DHPG to inhibit synaptic transmission. As shown in Fig. 4, B and C, both the initial and persistent depression induced by DHPG were dramatically enhanced in slices bathed in low-Ca²⁺ ACSF. Conversely, when slices were bathed in high-Ca²⁺ ACSF, both phases of the DHPG-induced depression were almost completely blocked (Fig. 4, B and C). Importantly, Eq. 1 predicts that if high extracellular Ca²⁺ inhibits the effects of DHPG by producing an increase in i_Ca²⁺ that compensates for a DHPG-induced inhibition of presynaptic Ca²⁺ signaling, then manipulations that decrease Ca²⁺ channel activity in the presence of high [Ca²⁺]_o should restore the ability of DHPG to depress synaptic transmission. Consistent with this we found that DHPG was able to induce a significant initial and persistent depression of synaptic transmission in slices bathed in high-Ca²⁺ ACSF that also contained the Ca²⁺ channel blocker Cd²⁺ (50 µM; Fig. 4D).

View larger version (36K): [in this window] [in a new window]   Fig. 4. Extracellular Ca2+ concentration strongly modulates the ability of DHPG to depress synaptic transmission. A: extracellular Ca2+ concentration modulates the effects of adenosine on synaptic transmission. In control experiments (, n = 6) done in the presence of normal levels of extracellular Ca2+ (2 mM), fEPSPs were reduced to 47.9 ± 6.4% of baseline during the last 5 min of a 10-min application of 25 µM adenosine (presence of adenosine in the recording chamber indicated by the bar). In slices bathed in a modified artificial cerebrospinal fluid (ACSF) containing 6 mM Ca2+ (high-Ca2+ ACSF), the inhibition of synaptic transmission by adenosine was significantly attenuated (, fEPSPs were reduced to 72 ± 6.7% of baseline, n = 7, P < 0.05 compared with control). In contrast, the inhibition of synaptic transmission induced by adenosine was enhanced in slices bathed in a modified ACSF containing 1.0 mM Ca2+ (low-Ca2+ ACSF). In these experiments () fEPSPs were reduced to 23.8 ± 2.9% of baseline during the last 5 min of the adenosine application, P < 0.05 compared with control. B: in slices bathed in high-Ca2+ ACSF, a 10-min application of 100 µM DHPG (indicated by ) had little effect on synaptic transmission (, n = 6). In these experiments, fEPSPs were 86.4 ± 8.8% of baseline in the presence of DHPG and were 92.8 ± 8.5% of baseline 30 min after DHPG washout. In contrast, decreasing [Ca2+]o to 1 mM strongly facilitated both the initial and persistent depression induced by DHPG (, n = 6). Here, fEPSPs were depressed to 0.44 ± 0.4% of baseline in the presence of DHPG and were 22.0 ± 12.6% of baseline 30 min after DHPG washout. Inset: fEPSPs recorded during baseline and at the end of the DHPG application in slices bathed in high-Ca2+ ACSF (left) and in low-Ca2+ ACSF (right). Calibration bars are 0.5 mV and 5.0 ms. C: summary of the effects of DHPG on synaptic transmission as a function of [Ca2+]o. *P < 0.05 compared with initial depression seen in control experiments ([Ca2+]o = 2 mM), **P < 0.05 compared with persistent depression seen in control experiments. D: the calcium channel blocker Cd2+ restores the induction of LTD by DHPG in high-Ca2+ ACSF. Slices were continuously bathed in a modified ACSF containing 6 mM Ca2+ and 50 µM CdCl2. In the presence of DHPG, fEPSPs were reduced to 48.2 ± 6.7% of baseline and were 56.1 ± 3.9% of baseline 30 min after DHPG washout (n = 5, P < 0.05 at both time points compare to pre-DHPG baseline).

Increasing the duration of presynaptic action potentials with the potassium channel blocker 4-aminopyridine (4-AP) attenuates the effects of voltage-sensitive Ca²⁺ channel blockers on transmitter release at excitatory synapses onto CA1 pyramidal cells (Wheeler et al. 1996). This occurs because the increase in P_o of presynaptic Ca²⁺ channels produced by prolonging action potentials with 4-AP presumably compensates for the decrease in N_tot produced by blocking specific Ca²⁺ channel subtypes (Wheeler et al. 1996). Similarly we found that the presynaptic inhibition of excitatory synaptic transmission induced by adenosine (25 µM) was significantly reduced in slices bathed in ACSF containing 100 µM 4-AP (in control experiments, a 10-min bath application of ACSF containing 25 µM adenosine reduced fEPSPs to 42 ± 8.5% of baseline, n = 7, while fEPSPs were reduced to only 83.1 ± 5% of baseline in 4-AP-treated slices, n = 5, P < 0.01 compared with control). Thus to further explore whether DHPG depresses synaptic transmission through effects on presynaptic Ca²⁺ signaling, we examined the effects of 4-AP on the ability of DHPG to inhibit synaptic transmission. Equation 1 predicts that increasing P_o by prolonging presynaptic action potentials with 4-AP should inhibit the depression induced by DHPG if DHPG depresses synaptic transmission by inhibiting presynaptic Ca²⁺ signaling. Consistent with this prediction, both the initial and persistent inhibition of synaptic transmission induced by DHPG were blocked in slices bathed in normal ACSF containing 100 µM 4-AP (Fig. 5A1). Moreover, Eq. 1 predicts that if 4-AP acts by producing an increase in P_o that compensates for a DHPG-induced inhibition of presynaptic Ca²⁺ signaling, then manipulations that decrease i_Ca²⁺ should offset the effects of 4-AP on P_o and restore the ability of DHPG to depress synaptic transmission. As predicted, DHPG was able to induce a significant initial and persistent depression of synaptic transmission in the presence of 4-AP when i_Ca²⁺ was reduced by lowering [Ca²⁺]_o to 0.5 mM (Fig. 5A2).

View larger version (33K): [in this window] [in a new window]   Fig. 5. The K+ channel blocker 4-aminopyridine (4-AP) suppresses the effects of DHPG on synaptic transmission while inhibition of postsynaptic potassium channels alone has no effect. A1: slices (with the CA3 region removed) were continuously bathed in either normal ACSF () or ACSF containing 100 µM 4-AP (). Following a 20-min period of baseline recording, a 10-min bath application of DHPG (100 µM, indicated by ) had little effect on synaptic transmission in 4-AP-treated slices. At the end of the DHPG application, fEPSPs were 102.9 ± 6.2% of baseline and were 117.9 ± 6.9% of baseline 30 min after DHPG washout (n = 8, P > 0.05 at both time points compared with pre-DHPG baseline). In control experiments (using slices with the CA3 region removed), fEPSPs were depressed to 34.7 ± 9.5% of baseline at the end of the DHPG application and were depressed to 57.9 ± 8.9% of baseline 30 min after DHPG washout (n = 6, P < 0.05 at both time points compared with pre-DHPG baseline). A2: DHPG induces a significant depression of synaptic transmission in the presence of 4-AP when [Ca2+]o is reduced. Slices (with the CA3 region removed) were continuously bathed in ACSF containing 100 µM 4-AP and 0.5 mM Ca2+. At the end of the DHPG application (indicated by ), fEPSPs were depressed to 55.9 ± 5.9% of baseline and were depressed to 70.9 ± 6.2% of baseline 30 min after DHPG washout (P < 0.05 at both time points compared with pre-DHPG baseline). B: blocking postsynaptic K+ channels alone does not suppress the DHPG-induced inhibition of synaptic transmission. In cells where the recording electrode contained both Cs+ (130 mM) and TEA (10 mM), EPSPs were depressed to 41.3 ± 4.9% of baseline at the end of a 10-min application of 100 µM DHPG and were 49.3 ± 7.3% of baseline 30 min after DHPG washout (, n = 9, P < 0.05 at both time points compared with pre-DHPG baseline). In contrast, the effects of DHPG on synaptic transmission were blocked in cells where 100 µM 4-AP was present in the ACSF and the recording electrode contained Cs+ and TEA (, n = 5). In these experiments, EPSPs were 90.0 ± 7.5% of baseline in the presence of DHPG and 111.2 ± 6.2% of baseline 30 min after DHPG washout. Inset: EPSPs (average of 3 responses) recorded during baseline and 30 min after DHPG washout in an experiment where only postsynaptic K+ channels were blocked with intracellular Cs+ and TEA (left) and in an experiment where Cs+ and TEA were present in the recording electrode and 4-AP was present in the bath (right). Calibration bars are 20 ms and 2.0 mV. C: cumulative probability plot showing the magnitude of the persistent depression of synaptic transmission induced by DHPG in each cell where EPSPs were recorded using a K+ gluconate-based electrode-filling solution (, average data shown in Fig. 1B) or a Cs+ gluconate-based electrode-filling solution containing TEA (, average data shown in B).

Although the results from our experiments with 4-AP and changes in extracellular Ca²⁺ are consistent with the hypothesis that activation of group I mGluRs with DHPG depresses synaptic transmission by modulating Ca²⁺ channel activity, they do not rule out the possibility that DHPG might instead regulate presynaptic Ca²⁺ signaling indirectly via an enhancement of presynaptic K⁺ channels and/or a decrease in the Ca²⁺ sensitivity of presynaptic proteins involved in transmitter release. Perhaps more importantly, these experiments do not directly demonstrate that 4-AP and changes in [Ca²⁺]_o modify the effects of DHPG by altering Ca²⁺ influx through pre- as opposed to postsynaptic Ca²⁺ channels. To address this second possibility, we examined whether blocking postsynaptic, but not presynaptic, K⁺ channels by introducing K⁺ channel blockers into postsynaptic CA1 pyramidal cells through a patch-clamp electrode could block the inhibition of synaptic transmission by DHPG. We did not attempt to use an electrode-filling solution containing 4-AP in these experiments because 4-AP is membrane permeant and thus could potentially diffuse out of the postsynaptic cell to affect presynaptic K⁺ channels. Instead, we examined whether the K⁺ channel blockers Cs⁺ or TEA might be useful for comparing how DHPG-induced LTD is affected when the same K⁺ channel blocker is bath applied to inhibit pre- and postsynaptic K⁺ channels or delivered intracellularly through the recording electrode to inhibit only postsynaptic K⁺ channels. To find concentrations of bath-applied Cs⁺ and TEA that had presynaptic effects similar to those produced by 100 µM 4-AP, we first determined the effects of 4-AP on the presynaptic fiber volley in slices where synaptic transmission was blocked by bathing slices in a modified ACSF containing 0 mM CaCl₂ and 10 mM MgSO₄. Under these conditions, a continuous application of 100 µM 4-AP induced a rapid increase in the duration of the presynaptic fiber volley that stabilized within 15-20 min after exposure to 4-AP. Forty minutes after slices were switched into ACSF containing 100 µM 4-AP the amplitude of the fiber volley was unchanged but its duration was increased to 160 ± 9% of baseline (n = 5). While 20-25 mM TEA produced a similar change in fiber volley duration (n = 4), we could not test the effects of bath-applied TEA on DHPG-induced LTD because we were unable to obtain stable baseline fEPSP recordings in slices bathed in normal ACSF containing TEA. We were also unable to determine whether inhibiting pre- and postsynaptic K⁺ channels with bath-applied Cs⁺ altered the effects of DHPG on synaptic transmission because we were unable to find a concentration of Cs⁺ that produced a stable change in the fiber volley similar to that induced by 4-AP. Intracellular Cs⁺ and TEA will, however, block several different types of postsynaptic K⁺ channels (Chen and Wong 1992; Velumian et al. 1993), including the A-type channels most likely inhibited by bath application of 4-AP (see Brown et al. 1990; Storm 1990 for reviews). Thus we examined the effects of DHPG on synaptic transmission in cells where the electrode-filling solution contained Cs⁺ and TEA. As can be seen from the results shown in Fig. 5B and the comparison shown in Fig. 5C, the DHPG-induced depression of synaptic transmission observed in cells where whole cell recordings were performed with electrodes containing Cs⁺ and TEA was the same as that seen in cells where the recording electrode was filled with a K⁺-based solution. In contrast, bath application of 4-AP strongly suppressed the DHPG-induced depression of synaptic transmission in cells where the recording electrode solution contained Cs⁺ and TEA (Fig. 5B). Thus while bath application of a low concentration of a K⁺ channel blocker, which will inhibit both pre- and postsynaptic K⁺ channels, strongly inhibits the depression of synaptic transmission by DHPG, selectively blocking postsynaptic K⁺ channels alone has no effect. This suggests that inhibition of presynaptic, rather than postsynaptic, K⁺ channels is responsible for the ability of 4-AP to oppose the DHPG-induced depression of synaptic transmission and supports the notion that pharmacological activation of group I mGluRs depresses synaptic transmission through presynaptic effects.

To further explore the potential role for presynaptic changes in LTD induced by DHPG, we performed two additional experiments. First, we examined paired-pulse facilitation (using an inter-pulse interval of 50 ms) before, during, and after application of 100 µM DHPG for 10 min. While previous studies have found effects of DHPG on paired-pulse facilitation in hippocampal neurons (Gereau and Conn 1995; Mannaioni et al. 2001), we observed no consistent effect of DHPG on paired-pulse facilitation of fEPSPs in either the presence of DHPG or following washout (n = 9, data not shown). Although this argues against a presynaptic locus for the depression induced by DHPG, studies at other synapses where mGluR agonists inhibit synaptic transmission through presynaptic effects have also failed to find consistent effects on paired-pulse facilitation (Barnes-Davies and Forsythe 1995). Thus as an additional test we examined whether NMDA, as well as AMPA, receptor-mediated synaptic currents were depressed by DHPG. If DHPG inhibits synaptic transmission through a presynaptic inhibition of transmitter release, then one prediction is that both the NMDA and AMPA receptor-mediated components of the postsynaptic excitatory currents recorded in CA1 pyramidal cells should be depressed. Consistent with this we found that a 5-min application of 100 µM DHPG induced a nearly identical and persistent depression of both AMPA and NMDA receptor-mediated EPSCs in voltage-clamped pyramidal cells (Fig. 6).

View larger version (24K): [in this window] [in a new window]   Fig. 6. DHPG induces LTD of both AMPA and NMDA receptor components of EPSCs in CA1 pyramidal cells. A 5-min application of 100 µM DHPG (indicated by ) induced a significant (P < 0.05 compared with baseline) persistent depression of both AMPA () and NMDA () components of EPSCs elicited by Schafer collateral fiber stimulation. Twenty-five to 30 min after the start of the DHPG application, AMPA receptor-mediated EPSCs were depressed to 58.6 ± 7.2% of baseline (n = 5) while NMDA receptor-mediated EPSCs were depressed to 63.8 ± 9.8% of baseline (n = 5). NMDA receptor-mediated EPSCs were recorded in the presence of 10 µM CNQX to block the AMPA receptor-mediated component of the EPSCs. Presynaptic fiber stimulation was delivered once every 20 s, and each point in the plot represents the average of 3 consecutive EPSCs. Insets: AMPA receptor-mediated (left) and NMDA receptor-mediated (right) EPSCs recorded during baseline and 25-30 min after DHPG application (smaller responses). Each trace is the average of 3 consecutive EPSCs. Calibration bars correspond to 25 ms and 25 pA.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Similar to previous observations in rat hippocampal slices (Palmer et al. 1997), our results show that DHPG, but not the broader spectrum mGluR agonist ACPD, induces a mGluR-dependent and NMDR-independent form of LTD of synaptic transmission in the CA1 and CA3 regions of the mouse hippocampus. Moreover, as has been observed in the rat hippocampus (Fitzjohn et al. 1998), the persistent depression of synaptic transmission induced by DHPG could be reversed by application of the mGluR antagonist LY341495, suggesting that activation of mGluR is not only required for the induction of DHPG-induced LTD but also for its expression. One possible interpretation of this second finding is that the persistent depression of synaptic transmission induced by DHPG is simply due to poor washout of the agonist from the slice. Two of our observations suggest that this is unlikely. First, after the expression of DHPG-induced LTD had been blocked with LY341495, LTD was reestablished when the antagonist was washed from the recording chamber. This suggests that DHPG-induced LTD is not simply due to activation of mGluRs by persistently bound DHPG because displacing DHPG from the receptors with LY341495 should facilitate DHPG washout and block both the expression and maintenance of the depression. Second, following co-application of DHPG and LY341495, synaptic transmission was not depressed when both compounds were washed from the recording chamber (Fig. 2) as would be expected if DHPG remained in the slice long after washing with drug-free ACSF. Together, these findings are more consistent with the recent suggestion that DHPG induces an "autopotentiation" of mGluRs (Merlin and Wong 1997; Schnabel et al. 2001). In this model, activation of mGluRs with DHPG induces a lasting change in mGluRs and/or downstream signaling pathways such that synaptically released glutamate can now persistently activate mGluRs and depress synaptic transmission. The mechanisms that might underlie such a process are currently unknown. Recent results showing that inhibitors of the calcium/calmodulin-dependent protein kinase CamKII (Schnabel et al. 1999b) and protein phosphatases (Schnabel et al. 2001) can modulate DHPG-induced LTD suggest, however, that changes in protein phosphorylation have an important role.

While the induction of mGluR-dependent forms of LTD by low-frequency patterns of synaptic stimulation occurs in the postsynaptic CA1 pyramidal cells, presynaptic changes in transmitter release are responsible, at least in part, for the expression of this form of LTD (Bolshakov and Siegelbaum 1994; Oliet et al. 1997). Consistent with these findings, our results suggest that the persistent depression of synaptic transmission induced by pharmacological activation of group I mGluRs with DHPG is dependent on postsynaptic GTP-dependent signaling pathways but is also strongly affected by experimental manipulations (altered [Ca²⁺]_o and 4-AP) that affect presynaptic Ca²⁺ influx. Although the dramatic effects of changes in [Ca²⁺]_o on the depression induced by DHPG are consistent with the notion that DHPG depresses transmission via effects on presynaptic Ca²⁺ signaling, changes in the concentration of extracellular divalent cations will also strongly effect neuronal excitability. Because under some conditions the ability of DHPG to depress synaptic transmission can be facilitated by manipulations that enhance neuronal excitability (Palmer et al. 1997), an alternative explanation is that elevating [Ca²⁺]_o inhibits the DHPG-induced depression of synaptic transmission by decreasing neuronal excitability while reducing [Ca²⁺]_o facilitates the effects of DHPG by enhancing excitability. Two of our findings suggest, however, that this is not the case. First, the addition of a low concentration of Cd²⁺ to the high-Ca²⁺ ACSF restored the ability of DHPG to induce a significant inhibition of synaptic transmission in high-Ca²⁺ ACSF even though Cd²⁺ should depress excitability even further. Second, the finding that 4-AP inhibits the ability of DHPG to depress synaptic transmission also seems more consistent with the interpretation that changes in [Ca²⁺]_o modify the ability of DHPG to depress synaptic transmission through effects on Ca²⁺ signaling. According to Eq. 1, if both the initial and persistent depression induced by DHPG arise from an inhibition of presynaptic Ca²⁺ signaling, then broadening the duration of presynaptic action potentials with 4-AP should have an effect similar to increasing i_Ca²⁺ by elevating [Ca²⁺]_o, i.e., the depression induced by DHPG should be inhibited in 4-AP-treated slices. On the other hand, if changing [Ca²⁺]_o regulates the ability of DHPG to depress synaptic transmission through changes in excitability, then the enhanced excitability due to 4-AP should have an effect similar to bathing slices in low-Ca²⁺ ACSF, i.e., the depression induced by DHPG should be enhanced in 4-AP-treated slices. Therefore 4-AP should have opposite effects on the DHPG-induced depression depending on whether changing [Ca²⁺]_o regulates the effects of DHPG on synaptic transmission by modulating Ca²⁺ signaling or by altering neuronal excitability. Our results showing that 4-AP blocks the depression induced by DHPG thus suggest that changes in excitability are unlikely to account for the effects of varying [Ca²⁺]_o on the DHPG-induced depression of synaptic transmission. Moreover, the ability of DHPG to depress synaptic transmission in 4-AP-treated slices when [Ca²⁺]_o is reduced to 0.5 mM supports the notion that 4-AP antagonizes the effects of DHPG by increasing Ca²⁺ channel P_o and also indicates that 4-AP does not inhibit the effects of DHPG under normal conditions by acting as a nonselective inhibitor of group I mGluR signaling.

While our results are consistent with the notion that DHPG depresses synaptic transmission by inhibiting presynaptic Ca²⁺ signaling, changes in [Ca²⁺]_o will also alter i_Ca²⁺ of both voltage-activated and ligand-gated Ca²⁺ channels (such as the NMDA receptor) in the postsynaptic cell. Likewise, bath application of 4-AP will also inhibit postsynaptic K⁺ channels and thus could enhance Ca²⁺ entry into the postsynaptic cell through voltage-activated as well as NMDA receptor ion channels. Our results therefore do not rule out the possibility that 4-AP and changes in [Ca²⁺]_o regulate the effects of DHPG by altering postsynaptic, rather than presynaptic, Ca²⁺ signaling. We found, however, that both the initial and persistent inhibition induced by DHPG were not affected in cells where the electrode-filling solution contained Cs⁺ and TEA at concentrations that should block postsynaptic K⁺ channels (Chen and Wong 1992; Velumian et al. 1993). Given the experimental limitations of the different K⁺ channel blockers used in our experiments, we were unable to directly compare how the ability of DHPG to depress synaptic transmission was altered when the same compound was used to block both pre- and postsynaptic K⁺ channels versus postsynaptic channels alone. It is important to note, however, that the combination of postsynaptic Cs⁺ and TEA used in our experiments will strongly block postsynaptic A-type K⁺ channels that are most sensitive to extracellular 4-AP (Brown et al. 1990; Chen and Wong 1992; Storm 1990; Velumian et al. 1993). Thus while inhibiting both pre- and postsynaptic K⁺ channels with bath applied 4-AP antagonizes the effects of DHPG, blocking postsynaptic K⁺ channels alone has no effect. This indicates that the ability of 4-AP to block the effects of DHPG on synaptic transmission can be attributed to an inhibition of presynaptic K⁺ channels, a result consistent with the hypothesis that DHPG inhibits excitatory synaptic transmission through effects on presynaptic calcium signaling. Because responses elicited by direct activation of postsynaptic NMDA receptors with exogenous agonists are enhanced by mGluRs activation (Aniksztejn et al. 1991; Fitzjohn et al. 1996; Mannaioni et al. 2001), our finding that DHPG induces a persistent depression of NMDA, as well as AMPA, receptor-mediated components of EPSCs in CA1 pyramidal cells also supports a presynaptic locus for the effects of DHPG on synaptic transmission.

In our experiments we found that both the initial depression of synaptic transmission seen in the presence of DHPG and the persistent depression of synaptic transmission that remained following DHPG washout were similarly affected by postsynaptic GDPS, extracellular 4-AP, and changes in [Ca²⁺]_o. Because our results suggest that the DHPG-induced inhibition of synaptic transmission may be due to presynaptic changes that occur following activation of postsynaptic mGluRs, it seems likely that both phases of the synaptic depression involve some form of retrograde signaling. One candidate retrograde messenger is arachidonic acid, which is thought to be involved in the induction of mGluR-dependent forms of LTD by synaptic stimulation (Bolshakov and Siegelbaum 1995) and can inhibit high-voltage-activated Ca²⁺ channels in hippocampal neurons (Keyser and Alger 1990). Endocannabinoids represent another possible candidate because activation of postsynaptic mGluRs in cerebellar Purkinje cells inhibits release from climbing fibers in a cannabinoid receptor-dependent manner (Maejima et al. 2001). Our results suggest, however, that if a diffusable retrograde messenger is involved in DHPG-induced LTD, it must act in a highly spatially restricted manner because loading single postsynaptic CA1 pyramidal cells with GDPS blocked the effects of bath applied DHPG. Interestingly, mGluR-dependent LTD induced by both synaptic stimulation and DHPG is inhibited when postsynaptic protein synthesis is blocked (Huber et al. 2000). Together with this finding our results suggest that a form of retrograde signaling involving postsynaptic protein synthesis-dependent processes that regulate presynaptic function may also be involved in mGluR-dependent forms of LTD.