Recurrent Mossy Fibers Establish Aberrant Kainate Receptor-Operated Synapses on Granule Cells from Epileptic Rats

Sprouting of mossy fibers in chronic epileptic rats

As in previous studies (Sutula et al., 1988; Mello et al., 1993; Represa et al., 1993; Okazaki et al., 1995; Buckmaster et al., 2002), an extensive sprouting of mossy fibers was observed in the dentate gyrus of chronic epileptic rats months after the pilocarpine injection and the inaugurating status epilepticus. Using Timm staining to reveal mossy fiber boutons, both the granule cell layer and the inner one-third of the molecular layer of the dentate gyrus were labeled in chronic epileptic rats (Fig. 1E), whereas only the hilar polymorph layer was labeled in controls (Fig. 1A). Because mossy fibers are glutamatergic (Henze et al., 2000), antibodies to VGLUT1, which specifically stain glutamatergic terminals (Fremeau et al., 2004), were also used to assess mossy fiber sprouting (Fig. 1B,F). Instead of the uniform staining with fine spots observed in control slices (Fig. 1B, inset), larger and brighter spots (Fig. 1F, inset) were observed in regions corresponding to recurrent mossy fiber terminal zones (i.e., the dentate granular and inner molecular layers) but not in other regions in which sprouting did not take place. According to this, we found a significant increase in VGLUT1 immunoreactivity in the inner molecular layer and the granular layer (by 152.4 ± 15.4%; unpaired t test; p = 0.001) but not in other regions (i.e., the outer molecular layer p = 0.76). The bigger spots displayed similar features than those observed in the CA3 stratum lucidum (data not shown) and may correspond to the newly formed mossy fiber synapses that are of larger size than other glutamatergic synapses and are highly enriched in clear, round synaptic vesicles (Represa et al., 1993). Therefore, the dentate gyrus from chronic epileptic rats displays sprouting of glutamatergic mossy fibers in our experimental conditions.

Kainate receptor-mediated synaptic events in granule cells from chronic epileptic but not control rats

In control GCs, low-frequency stimulations (0.033 Hz) in the inner one-third of the molecular layer evoked a synaptic current with a rapid decay in the continuous presence of antagonists of GABA_A (10 μm bicuculline) and NMDA (40 μm d-APV) receptors (Fig. 1C, left). This current was mediated by AMPA receptors, because it was completely blocked by the AMPAR antagonist GYKI 52466 (Fig. 1C, middle) (100 μm; n = 5), even when repetitive stimulations were applied (Fig. 1C, right) (10 stimuli at 30 Hz). In dentate GCs from chronic epileptic rats (n = 5), low-frequency stimulations (0.033 Hz) of recurrent mossy fibers in the inner one-third of the molecular layer evoked a synaptic current with a fall composed of both a rapid and a slow component (Fig. 1G, left) (n = 5). Application of GYKI 52466 (100 μm) blocked the rapid component but not the slow component (Fig. 1G, middle). The slow component was mediated by KARs, because it was subsequently blocked by the mixed AMPAR/KAR antagonist CNQX (Fig. 1G, right) (50 μm). Therefore, stimulation in the inner one-third of the dentate molecular layer evoked KAR-mediated synaptic responses in dentate granule cells from epileptic rats but not control rats.

We next tested whether KARs can be activated by the spontaneous release of glutamate (Cossart et al., 2002) in dentate GCs from epileptic rats. Dentate GCs were recorded at the reversal potential for GABAergic currents (V_h =–70 mV), a potential at which NMDAR-mediated currents are also abolished by Mg²⁺ block (Ascher and Nowak, 1988). In dentate GCs from epileptic rats, the frequency of spontaneous EPSCs was significantly higher than that in control rats (compare Fig. 1D,H) (1.65 ± 0.54 Hz, n = 18 and 0.41 ± 0.06 Hz, n = 17, respectively; Mann–Whitney; p = 0.021) as reported previously (Wuarin and Dudek, 2001). Visual inspection of EPSCs revealed the presence of events with fast and slow kinetics in GCs from epileptic rats, whereas only fast events were observed in control cells (compare Fig. 1D,H). The fast events recorded in control GCs were mediated by AMPARs, because they were all blocked by GYKI 52466 (Fig. 1D) (100 μm; n = 5). The slow EPSCs recorded in GCs from epileptic rats were mediated by KARs, because they were resistant to GYKI 52466 (Fig. 1H) (100 μm; n = 5) and fully abolished by the mixed AMPAR/KAR antagonist CNQX (50 μm; n = 5; data not shown). In control GCs, the nonobservation of slow EPSC_KA could result from the low frequency of spontaneous synaptic activity in these cells. To address this, we artificially increased the spontaneous synaptic activity in control cells using 4-AP (50–100 μm), a potassium channel antagonist, in the continuous presence of bicuculline (10 μm) and d-APV (40 μm). In the presence of 4-AP, the frequency of spontaneous EPSCs recorded in control GCs was drastically increased (Fig. 2A–C) (from 0.41 ± 0.08 to 1.61 ± 0.32 Hz; n = 5; Mann–Whitney; p = 0.008) and was not significantly different from that observed in GCs from epileptic rats (Fig. 2C) (n = 18; Mann–Whitney; p = 0.146). In these conditions, all non-NMDAR-mediated EPSCs were fully abolished by AMPAR antagonists NBQX (Fig. 2A,B)(1 μm; n = 3) (Bureau et al., 1999; Cossart et al., 2002) or GYKI 52466 (100 μm; n = 2). Thus, KARs contribute to the spontaneous and evoked synaptic transmission in granule cells from chronic epileptic rats but not control rats.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Mossy fiber sprouting and EPSC_KA in granule cells from epileptic rats. A, E, Timm staining (dark brown; counterstained with cresyl violet) of mossy fibers of control (A) and epileptic (E) rat sections after electrophysiological recordings. In chronic epileptic rats, mossy fiber collaterals invade the granule cell layer (g) and the inner third of the molecular layer (iml). Scale bars, 20μm. B, F, VGLUT1 immunostaining of sections from control (B) and epileptic (F) rats. In epileptic rats, the staining dramatically increases in the inner molecular layer (iml) and granule cell layer, in good correlation with the sprouting of mossy fibers. In the inner molecular layer, the type of staining is modified, as large spots (characteristic of the staining of large mossy fiber synaptic boutons) have replaced the fine speckled staining observed in the control (enlarged in the insets). Scale bars, 50μm; insets, 10μm. C, G, Averages of 30 individual EPSCs to stimulations (0.033 Hz) in the inner one-third of the molecular layer of the dentate gyrus recorded in the presence of 10μm bicuculline (Bicu) and 40μm d-APV (APV). In control granule cells (C), the EPSC evoked in control conditions (left) was completely blocked by AMPA receptor antagonist GYKI 52466(GYKI; 100 μm; middle), even when repetitive stimulations were applied (10 stimuli at 30 Hz; stim 30 Hz; right). In dentate GCs from epileptic rats (G), the EPSC evoked in control conditions displayed a rapid and a slow decay (left). Application of GYKI 52466 (GYKI; 100 μm) blocked the rapid but not the slow component (middle); the slow component was subsequently abolished by the mixed AMPAR/KAR antagonist CNQX (50 μm; right). D, H, Representative sEPSC recordings (V_h =–70 mV in this and subsequent figures) from a control granule cell (D) and a granule cell from pilocarpine-treated rat (H), showing that the non-NMDAR-mediated synaptic transmission is enhanced in epileptic tissue. Note that in control tissue, all of the synaptic events are fast (filled circle; D) and mediated by AMPARs, because they are fully abolished by 100 μm GYKI 52466 (D). In contrast, in the epileptic tissue, the non-NMDAR-mediated synaptic transmission additionally includes slow events (asterisk; H) that are resistant to 100 μm GYKI 52466 (H).

To determine whether the contribution of KARs to spontaneous synaptic transmission in GCs occurred before or after the onset of chronic epilepsy, we recorded spontaneous non-NMDAR-mediated EPSCs in 4- to 5-d-poststatus epilepticus rats that did not express behavioral seizures and showed no evidence of mossy fiber sprouting using Timm staining (data not shown). The frequency of spontaneous EPSCs recorded in GCs from 4- to 5-d-poststatus epilepticus rats was not significantly different from that recorded in control GCs (0.45 ± 0.17 Hz, n = 5 and 0.41 ± 0.06 Hz, n = 17, respectively; Mann–Whitney; p = 1.00) (supplemental Fig. S1A,C, available at www.jneurosci.org as supplemental material), as reported previously (Wuarin and Dudek, 2001). Visual inspection of EPSCs revealed the presence of events with fast kinetics only that were all blocked by the AMPAR antagonist GYKI 52466 (100 μm; n = 5; data not shown). As in control GCs, the nonobservation of slow EPSC_KA in GCs from 4- to 5-d-poststatus epilepticus rats could result from the low frequency of spontaneous synaptic activity in these cells. Therefore, we increased the spontaneous synaptic activity in GCs from 4- to 5-d-poststatus epilepticus rats using 4-AP (100 μm) in the continuous presence of bicuculline (10 μm) and d-APV (40 μm). In the presence of 4-AP, the frequency of spontaneous EPSCs was drastically increased (from 0.45 ± 0.17 to 4.57 ± 1.22 Hz; n = 5; paired t test; p = 0.028) (supplemental Fig. S1 A, C, available at www.jneurosci.org as supplemental material). In these conditions, all non-NMDAR-mediated EPSCs were fully abolished by AMPAR antagonists NBQX (1 μm; n = 3) (supplemental Fig. S1 A, B, available at www.jneurosci.org as supplemental material) or GYKI 52466 (100 μm; n = 2). In conclusion, KARs contribute to synaptic transmission in granule cells from chronic epileptic rats but not control rats or 4- to 5-d-poststatus epilepticus rats.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Control granule cells do not generate EPSC_KA, even in conditions of increased spontaneous activity. A, Spontaneous EPSC recordings in the presence of bicuculline (Bicu; 10 μm) and d-APV (40 μm) in a control granule cell. Note that application of 4-AP (50μm) strongly increases the frequency of EPSCs, which are all mediated by AMPARs, because they are fully abolished by 1 μm NBQX. B, Histogram plot of the number of spontaneous EPSCs versus time before and during bath application of 50 μm 4-AP in a representative control granule cell. The frequency of spontaneous EPSCs is strongly increased during 4-AP application. Subsequent application of 1 μm NBQX fully abolished all of the spontaneous EPSCs. C, Bar graphs of the mean frequency of spontaneous EPSCs recorded in control granule cells (in the presence of bicuculline and d-APV; n = 5) in the absence (Cont) and in the presence of 4-AP (50 μm). Note that in the presence of 4-AP, the mean frequency of spontaneous EPSCs is significantly increased (**p < 0.01) and similar to that of spontaneous EPSCs recorded in granule cells from pilocarpine-treated rats (n = 18; p > 0.05).

Kinetics of mEPSC_AMPA and mEPSC_KA in granule cells from epileptic rats

We next characterized the kinetics of pharmacologically isolated mEPSC_AMPA and mEPSC_KA to subsequently determine their contribution to synaptic activity in the absence of any antagonist (Cossart et al., 2002). Miniature EPSCs (mEPSCs) were recorded in the presence of TTX (1 μm), bicuculline (10 μm), and d-APV (40 μm) to isolate non-NMDAR-mediated glutamatergic events. There was no significant difference between the frequencies of miniature glutamatergic events recorded in dentate GCs from control rats and those from chronic epileptic rats (0.25 ± 0.09 Hz for controls, n = 14 and 0.27 ± 0.06 Hz for epileptic rats, n = 12; Mann–Whitney; p = 0.487). However, we observed a significant shift toward slower values when comparing the cumulative probability plots of decay times of all mEPSCs recorded in GCs from epileptic and control rats (Fig. 3D) (χ² test; p < 0.001). In GCs from control rats (n = 14), all miniature events displayed a fast kinetics (Fig. 3A) and clustered in one area of the graph in scatter plots of 10–90% rise times versus decay times (Fig. 3B). The histogram of the decay times of miniature events revealed that fast events belonged to one population, because their distribution was best fitted by one Gaussian curve (Fig. 3C, left). All of these events were mediated by AMPARs, because they were fully blocked by NBQX (Fig. 3A) (1 μm; n = 9) or GYKI 52466 (100 μm; n = 5; mEPSC_AMPA; r = 1.29 ± 0.07 ms; τ = 3.26 ± 0.18 ms; n = 14).

In GCs from epileptic rats (n = 12), we observed slow, miniature events (mEPSC_slow) in addition to the fast events (mEPSC_fast) (Fig. 3A). mEPSCs clustered within two separated areas of the graph in scatter plots of 10–90% rise times versus decay times (Fig. 3B). The histogram of the decays of miniature events revealed that mEPSC_fast and mEPSC_slow belonged to two distinct populations, because it was best fitted by two statistically different Gaussian curves (Fig. 3C, middle) (Kolmogorov–Smirnov; p < 0.005). Bath applications of AMPAR antagonists (1 μm NBQX, n = 6; or 100 μm GYKI 52466, n = 5) selectively suppressed the fast events (Fig. 3A) as shown by the histogram of the decay times (compare Fig. 3C, middle and right) and the cumulative probability plot of the decay times (Fig. 3D) as well as the plot of 10–90% rise times versus decay times (Fig. 3B). The miniature events that remained in the presence of AMPAR antagonists were mediated by KARs, because they were fully abolished by the mixed AMPAR/KAR antagonist CNQX (Fig. 3A) (50 μm; mEPSC_KA, r = 2.53 ± 0.28; τ = 9.83 ± 0.82; n = 11). mEPSC_KA and mEPSC_slow (recorded in the absence of the AMPAR antagonist) represented the same population of events, because the Gaussian curve fitting the mEPSC_KA distribution and the one fitting mEPSC_slow were not significantly different (compare Fig. 3C, right and middle) (Kolmogorov–Smirnov; p > 0.1). In keeping with this, the group of events obtained when plotting 10–90% rise times versus decay times of mEPSC_KA overlapped that of mEPSC_slow but not that of mEPSC_fast (Fig. 3B). To pharmacologically isolate mEPSC_AMPA in GCs from epileptic rats, we used bath application of SYM 2081 (n = 5), a compound that selectively desensitizes KARs (DeVries and Schwartz, 1999; Li et al., 1999). SYM 2081 (10 μm) blocked the slow events (Fig. 4A) as shown by the histogram (Fig. 4C) and the cumulative probability plot of the decay times of miniature events (Fig. 4D) as well as the plot of 10–90% rise times versus decay times (Fig. 4B). In the presence of SYM 2081, the remaining fast miniature events were mediated by AMPARs, because they were fully abolished by GYKI 52466 (Fig. 4A) (100 μm; mEPSC_AMPA; r = 1.20 ± 0.28; τ = 3.15 ± 0.31; n = 5). mEPSC_AMPA and mEPSC_fast (recorded in the absence of the KAR antagonist) represented the same population of events, because the Gaussian curve fitting the mEPSC_AMPA distribution and the one fitting mEPSC_fast were not significantly different (compare Fig. 4C, left and right) (Kolmogorov–Smirnov; p > 0.1). In keeping with this, the group of events obtained when plotting 10–90% rise times versus decay times of mEPSC_AMPA overlapped that of mEPSC_fast but not that of mEPSC_slow (Fig. 4B). Furthermore, the kinetics of mEPSC_AMPA recorded in control GCs and those pharmacologically isolated in GCs from epileptic rats were not significantly different because of the following: (1) the Gaussian curve fitting the distribution of mEPSC_AMPA in the presence of SYM 2081 in GCs of epileptic rats and the one fitting mEPSC_AMPA in control GCs were not significantly different (compare Figs. 3C, left, 4C, right) (Kolmogorov–Smirnov; p > 0.1); (2) in cumulative probability plots, the distribution of mEPSCs recorded in GCs from epileptic rats in the presence of SYM 2081 was not significantly different from that recorded in control GCs (Fig. 4D) (χ² test; p > 0.1); and (3) they had similar rise and decay time mean values (unpaired t test; p = 0.776 for 10–90% rise time and p = 0.667 for decay time) (supplemental table, available at www.jneurosci.org as supplemental material).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

mEPSC_KA can be pharmacologically isolated in the presence of AMPA receptor antagonists in granule cells from epileptic rats. A, mEPSC recordings in the presence of 1μm TTX, 10μm bicuculline (Bicu), and 40μm d-APV in a granule cell from control and epileptic rats. In the control cell, mEPSCs have only fast kinetics. In the cell from an epileptic rat, fast (filled circles) and slow (asterisk) events can be recorded. Fast events are mediated by AMPARs, because they are blocked by 1 μm NBQX. Slow mEPSCs are mediated by KARs, because they are NBQX resistant and abolished by 50μm CNQX. B, Scatter plots of mEPSC 10–90% rise times versus decay times in a granule cell from a control and epileptic rat. Note that in the granule cell from an epileptic rat, events cluster in two separate areas of the graph, in contrast to the control cell. In the presence of 1 μm NBQX, only events with a slow time course are present in the granule cell from an epileptic rat. C, Distribution of decay times of all mEPSCs in granule cells from control (left) and epileptic (middle) rats. In granule cells from epileptic rats, the distribution is fitted with two Gaussian curves (dotted lines; mean, 3.2 and 8.9 ms, respectively) in contrast to control neurons (mean, 3.1 ms). In granule cells from epileptic rats, only mEPSCs with a slow decay time (346 events recorded in 11 cells; right) are resistant to AMPAR antagonists (1μm NBQX or 100 μm GYKI 52466); the distribution is now fitted by one Gaussian curve (mean, 9 ms). D, Cumulative probability plot of the decay time constant of all mEPSCs in granule cells from control (cross) and epileptic (filled circle) rats. In granule cells from epileptic rats, the decay time distribution is significantly shifted toward slower values (χ² test; p < 0.001). The addition of AMPAR antagonists (1μm NBQX or 100μm GYKI 52466) further shifted the curve toward slower values (open circle; χ² test; p < 0.01) in accordance with the blockade of fast EPSC_AMPA.

We conclude that in GCs from epileptic rats but not control rats, the quantal release of glutamate generates slow mEPSC_KA in addition to fast mEPSC_AMPA.

Kainate receptors provide half of the spontaneous and miniature glutamatergic currents in granule cells from epileptic rats

To quantify the contribution of AMPARs and KARs to synaptic transmission in dentate GCs from epileptic rats, we statistically determined limits for decay times to segregate EPSC_AMPA and EPSC_KA in the absence of AMPAR or KAR antagonists (see Materials and Methods). The mean rise and decay time values of EPSC_AMPA and EPSC_KA selected using these decay time limits were not significantly different from those of pharmacologically isolated events (n = 12; paired t test; mEPSC_AMPA, p = 0.150 for 10–90% rise time and p = 0.646 for decay time; mEPSC_KA, p = 0.428 for 10–90% rise time and p = 0.610 for decay time) (supplemental table, available at www.jneurosci.org as supplemental material).

Using the kinetic limits, we found that the frequencies were 0.15 ± 0.05 and 0.12 ± 0.02 Hz for mEPSC_KA and mEPSC_AMPA, respectively (n = 12). Thus, mEPSC_KA represented half of the miniature non-NMDA glutamatergic synaptic events (50.8 ± 6.63%; n = 12). We extended this analysis by calculating the mean charge carried by KARs and AMPARs in digitally averaged mEPSC_KA and mEPSC_AMPA. We observed that, as reported previously in other cell types (Frerking and Nicoll, 2000; Cossart et al., 2002), the mean charge transferred during EPSC_KA was significantly higher than that during EPSC_AMPA (n = 12; Mann–Whitney; p = 0.014). Considering the frequency of synaptic events and the charge transferred, we estimated that KARs provided as much as half of the total current mediated by non-NMDARs in dentate GCs from pilocarpine-treated rats (56.2 ± 5.98%; n = 12). We next determined the contribution of KARs to the spontaneous activity. The same limits for decay times were used to segregate between spontaneous EPSC_KA and EPSC_AMPA. We found that a significant part of non-NMDAR-mediated spontaneous events recorded in GCs from pilocarpine-treated rats was mediated by KARs (45.48 ± 6.81%; n = 18). Considering the frequency and the charge transferred, we estimated that half (52.0 ± 6.73%; n = 18) of the non-NMDAR-mediated spontaneous glutamatergic drive on dentate GCs was mediated by KARs in pilocarpine-treated rats.

Therefore, KARs are major contributors to miniature and spontaneous synaptic transmission in granule cells from chronic epileptic rats. We then determined whether EPSC_KAs were generated by recurrent mossy fibers in these cells.

Single EPSC_KA evoked by minimal stimulation shows pharmacological and physiological features of recurrent mossy fiber synapses in granule cells from epileptic rats

Minimal stimulations were performed in the inner one-third of the molecular layer to stimulate single recurrent mossy fibers in the presence of a high-divalent cation concentration (8 mm Ca²⁺), 10 μm bicuculline, and 40 μm d-APV. All-or-none EPSCs (Fig. 5A) with slow kinetics similar to mEPSC_KA could be evoked in GCs from epileptic rats (supplemental table, available at www.jneurosci.org as supplemental material). These EPSCs were mediated by KARs, because they were resistant to applications of NBQX (1 μm) but fully blocked by CNQX (Fig. 5A,B) (50 μm; n = 5). Fast EPSC_AMPA (fully blocked by 1 μm NBQX) could also be evoked (n = 2; data not shown).

To ensure that minimally evoked EPSC_KAs were specifically generated by mossy fibers, we relied on the well established sensitivity of mossy fiber EPSCs to the metabotropic GluR2/3 (mGluR2/3) agonist DCGIV (Kamiya et al., 1996). In pilocarpine-treated animals, this agent also reduced synaptic transmission at recurrent mossy fiber synapses (Feng et al., 2003). Applications of DCGIV (1 μm) enhanced significantly the failure rate of the minimal eEPSC_KA (Fig. 5C,D) (from 64.5 ± 3.12% in control to 92.6 ± 1.39% in the presence of DCGIV; washout, 69.7 ± 3.89%; n = 6; paired t test; p < 0.001). In contrast, the amplitude of the eEPSC_KA was not changed (Fig. 5C) (control, 4.04 ± 0.54 pA; DCGIV, 3.71 ± 0.19 pA; n = 6; paired t test; p = 0.549). We also relied on the selective form of frequency facilitation expressed both by control and recurrent mossy fiber EPSCs (Salin et al., 1996; Feng et al., 2003). We found that increasing the frequency of stimulation from 0.2 to 1 Hz reduced the failure rate of the minimal eEPSC_KA from 78.8 ± 7.96% at 0.2 Hz to 51.2 ± 7.68% at 1 Hz (Fig. 5E,F) (n = 5; paired t test; p = 0.037). Enhancing further the frequency of stimulation from 1 to 10 Hz did not decrease further the failure rate (Fig. 5E,F) (failure rate at 10 Hz, 51.9 ± 11.2%; paired t test; p = 0.928). Therefore, EPSC_KA evoked in the granule cells from epileptic rats displayed specific pharmacological and physiological features of mossy fiber synapses.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

mEPSC_AMPA can be pharmacologically isolated in the presence of a kainate receptor blocker in granule cells from epilepticrats.A, mEPSC recordings in the presence of 1 μm TTX, 10 μm bicuculline (Bicu), and 40 μm d-APV from a granule cell from epileptic rats. Slow events (asterisk) are mediated by KARs, because they are blocked by 10μm SYM 2081. Fast events (filled circle) are mediated by AMPARs, because they are SYM 2081 resistant and abolished by 100 μm GYKI 52466. B, Scatter plots of mEPSC, 10–90% rise times versus decay times in a granule cell from an epileptic rat. Note that in the presence of 10 μm SYM 2081, only events with a fast time course are present (right). C, Distribution of decay time constant of all mEPSCs in granule cells from epileptic rats in the absence (left) and in the presence of 10 μm SYM 2081 (right). Note that in the presence of 10 μm SYM 2081, the distribution of the decay time constant of miniature EPSCs is fitted by one Gaussian curve (mean, 3.3 ms). D, Cumulative probability plot of the decay time constant of all mEPSCs recorded in granule cells from control (cross) and epileptic (filled circle) rats. Note that in the presence of 10 μm SYM 2081, the cumulative probability plot is shifted toward faster values (233 events recorded in 5 neurons; open circle; χ² test; p < 0.001) and is superimposed to that of control cells (χ² test; p > 0.05).

Recurrent mossy fibers generate both EPSC_KA and EPSC_AMPA in granule cells from epileptic rats

To further identify mEPSCs generated by recurrent mossy fibers in GCs from epileptic rats, we tested the effect of FSK, an adenylyl cyclase activator, which has been shown to selectively enhance synaptic transmission at mossy fiber synapses (Weisskopf et al., 1994). Miniature non-NMDAR-mediated EPSCs were recorded in the presence of TTX (1 μm), bicuculline (10 μm), and d-APV (40 μm). Bath application of FSK (10 μm) induced a gradual and long-lasting increase in the frequency of mEPSCs to 380 ± 59% of control (Fig. 6A,B) (paired t test; p = 0.016; n = 5) without significant change in amplitude (see Fig. 6D for the distribution) (116 ± 18% of control; paired t test; p = 0.508; n = 5). The frequency analysis of mEPSC_KA and mEPSC_AMPA using our kinetic limits showed that both were significantly enhanced 20 min after FSK application (Fig. 6A,C) (to 433 ± 75% of control, paired t test, p = 0.035 for mEPSC_KA; to 350 ± 48% of control, paired t test, p = 0.009 for mEPSC_AMPA; n = 5 for each). Thus, mEPSC_KAs are highly sensitive to FSK application, and this further supports the mossy fiber origin of these events. These results also suggest that mossy fibers generate EPSC_AMPA, as reported previously (Okazaki and Nadler, 2001).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Single EPSC_KA are evoked by minimal stimulation of mossy fibers in granule cells from epileptic rats. A, Left, Representative superimposed recordings of minimally evoked unitary EPSCs (eEPSCs; n = 6) in granule cells from epileptic rats. Minimal stimulations (arrow) were performed in 8 mm Ca²⁺, 10 μm bicuculline (Bicu), and 40 μm d-APV. Unitary events evoked in these conditions are mediated by KARs (eEPSC_KA), because they are resistant to 1 μm NBQX (middle) and are fully abolished by 50 μm CNQX (right). B, Bar graphs of averaged values of rise times (left) and decay times (right) show that the time course of unitary slow events and NBQX (1μm)-resistant events recorded in dentate granule cells of epileptic rats (n = 5) are not significantly different (paired t test; p = 0.480 for 10–90% rise times and p = 0.180 for decay times). C, The plot of amplitude of unitary eEPSC_KA versus time shows that 1 μm DCGIV (mGluR2/3 agonist) does not change their amplitude but increases their failure rate (see also D). D, Bar graphs of the mean failure rate of unitary eEPSC_KA calculated before (Cont.), during, and after (wash) application of 1 μm DCGIV. Note that the mGluR2/3 agonist significantly increases the failure rate of eEPSC_KA (***p < 0.001; n = 6). E, F, Consecutive traces of unitary eEPSC_KA (n = 10) minimally stimulated at 0.2, 1, and 10 Hz (E) and bar graphs of the mean failure rate of these events calculated for each frequency (F; 50 stimulations for each frequency in 6 cells) show that increasing the stimulation rate from 0.2 to 1 Hz resulted in a significant reduction of failures (*p < 0.05); increasing the frequency of stimulation from 1 to 10 Hz does not further decrease the failure rate (p > 0.05).

To confirm that recurrent mossy fibers generated EPSC_AMPA in addition to EPSC_KA, we selectively stimulated dentate GCs using focal pressure application of the nonhydrolysable glutamate agonist KA to the molecular layer while recording from a distant GC (Fig. 7A) (Wuarin and Dudek, 1996). Methylene blue was routinely added to the agonist solution to visualize the flow during pressure application and evaluate the spatial extent of the stimulation. Distribution of the ink typically indicated a diffusion of ∼150 μm from the tip of the pipette. We also used a high divalent cation-containing medium (Ca²⁺ 8 mm), which blocked polysynaptic activities (Berry and Pentreath, 1976), to avoid unspecific spread of activity. Pressure application of KA (30 μm; 20–50 ms duration) directly on the recorded GC in the presence of d-APV (40 μm) and bicuculline (10 μm) generated an inward current that lasted for several seconds (typically ∼60 s) (Fig. 7A). Focal applications of KA with a micropipette placed in the molecular layer of the dentate gyrus from a pilocarpine-treated rat, but at a distance of ≥300 μm from the recorded cell, increased the frequency of EPSCs (in 7 of 28 GCs) without underlying current (Fig. 7B). This effect lasted between 30 and 60 s (i.e., in the same range as the current evoked when applying KA directly on the recorded cell) (Fig. 7A). The frequency analysis of EPSC_AMPA and EPSC_KA showed that both were significantly enhanced (Fig. 7C,D) (to 264 ± 58% of control, paired t test, p = 0.04 for EPSC_AMPA; to 244 ± 35% of control, paired t test, p = 0.004 for EPSC_KA; n = 7 for each). Similar stimulations in the molecular layer of the dentate gyrus from control rats failed to change the frequency of EPSCs in distant recorded dentate GCs (n = 20; data not shown) as described previously (Wuarin and Dudek, 1996). Therefore, the recurrent mossy fibers generated EPSC_AMPA in addition to EPSC_KA in dentate granule cells from epileptic rats.

To test the specificity of these responses, we determined whether spontaneous EPSC_KAs could also be generated by mossy cells that may survive the pilocarpine-induced status epilepticus (Scharfman et al., 2001). Using an antibody against GluR2/3 to label mossy cells in the hilar region (Leranth et al., 1996), we found that some glutamatergic hilar mossy cells survived the status epilepticus (33–47%) (supplemental Fig. S2A, available at www.jneurosci.org as supplemental material) as described previously (Scharfman et al., 2001). Using the same focal stimulation protocol in the hilus (supplemental Fig. S2B, available at www.jneurosci.org as supplemental material), we observed an increase in the frequency of EPSCs in GCs (n = 7 of 15) (supplemental Fig. S2C, available at www.jneurosci.org as supplemental material) during 30–60 s. However, only EPSC_AMPA frequency was significantly increased (to 533 ± 211% of control; n = 7; paired t test; p = 0.024); the frequency of EPSC_KA was not changed (107 ± 23% of control; n = 7; paired t test; p = 0.855) (supplemental Fig. S2D,E, available at www.jneurosci.org as supplemental material). As expected, in control hippocampal slices, similar stimulations in the hilus enhanced the frequency of EPSC_AMPA to 439 ± 136% of control (n = 3 of 4 GCs) (data not shown).

Altogether, these experiments showed that recurrent mossy fibers generate both EPSC_KAs and EPSC_AMPAs in dentate granule cells from pilocarpine-treated rats.

Blockade of kainate receptors reduces synchronized network-driven activities in the dentate gyrus from epileptic rats

Because mossy fiber sprouting leads to the formation of a recurrent excitatory circuit between GCs in epileptic rats (Tauck and Nadler, 1985; Wuarin and Dudek, 1996; Molnar and Nadler, 1999; Buckmaster et al., 2002; Scharfman et al., 2003) that support synchronized burst discharges (Patrylo and Dudek, 1998; Hardison et al., 2000; Okazaki and Nadler, 2001), we then determined whether blockade of KARs could reduce recurrent mossy fiber network-driven activities. To test this hypothesis, synchronized burst discharges were evoked by antidromic stimulation of the mossy fiber pathway in the CA3b area in high potassium containing ASCF (6 mm) in the continuous presence of bicuculline (Fig. 8A) (10 μm) (Patrylo and Dudek, 1998; Hardison et al., 2000) while recording local field potentials in the dentate granular layer. In these conditions, we were able to evoke burst activities in 9 of 10 slices from epileptic rats but in none of the nine control slices tested (Fig. 8B). We used the coastline burst index to quantify the involvement of KARs in burst discharges (Dingledine et al., 1986). Desensitization of KARs using SYM 2081 (10 μm) significantly reduced the coastline burst index in eight of nine tested slices to 56.8 ± 4.58% of control condition (Fig. 8B) (n = 682 bursts; Wilcoxon; p = 0.008) without affecting the initial population spike (100 ± 6.08% of control; n = 9; Wilcoxon; p = 0.82). Thus, KARs are involved in synchronized network-driven activities supported by recurrent mossy fibers in the dentate gyrus from epileptic rats.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

FSK induces a long-lasting enhancement of the frequency of mEPSC_KA and mEPSC_AMPA in granule cells from epileptic rats. A, mEPSC recordings in the presence of 1 μm TTX, 10μm bicuculline (Bicu), and 40μm d-APV in a granule cell from an epileptic rat in the absence (left) and 20 min after the application of 10μm FSK (right). Note that in the presence of FSK, the frequency of both mEPSC_AMPA (filled circle) and mEPSC_KA (asterisk) is increased. B, Time course of the FSK effect on the mean frequency of miniature events (n = 5). C, Bar graphs of the mean frequency of mEPSC_AMPA and mEPSC_KA recorded before (Cont.) and 20 min after the application of FSK. The frequency of mEPSC_KA and mEPSC_AMPA is significantly increased (n = 5; *p < 0.05; **p < 0.01). D, Cumulative probability plots of amplitudes of miniature EPSCs before (solid line) and 20 min after (dotted line) FSK application. Note that there is no significant difference in the amplitudes of mEPSCs before and 20 min after FSK application.

Mossy fiber origin of EPSC_KA

In the control hippocampus, mossy fiber synapses display both presynaptic- and postsynaptic-specific features (Henze et al., 2000). On the presynaptic side, these synapses are downregulated by the activation of mGluR2/3 receptors (Kamiya et al., 1996) and present a specific form of frequency facilitation (Salin et al., 1996) via the activation of presynaptic KARs (Schmitz et al., 2001); on the postsynaptic side, mossy fiber synapses onto CA3 pyramidal cells selectively generate EPSC_KA (Castillo et al., 1997; Cossart et al., 2002). Previous studies have shown that presynaptic features of mossy fibers are preserved after sprouting in the epileptic hippocampus (Feng et al., 2003). We now report that, in addition to these presynaptic features, recurrent mossy fibers also conserve their postsynaptic specificity, because they generate EPSC_KA: (1)

EPSC_KA are recorded in chronic epileptic rats that showed strong mossy fiber sprouting but not in control or 4- to 5-d-poststatus epilepticus rats that showed no mossy fiber sprouting; (2) in chronic epileptic rats, EPSC_KA can be evoked by minimal stimulation in the inner molecular layer of the dentate gyrus, the terminal zone of recurrent mossy fibers; (3) these unitary EPSC_KAs show pharmacological and physiological features specific of mossy fiber synapses such as sensitivity to DCGIV and forskolin and frequency-dependent facilitation; (4) EPSC_KA can be generated by the focal stimulation of distant GCs but not hilar mossy cells; and (5) blockade of KARs reduce recurrent mossy fiber network-driven activities. These results suggest that the establishment of new KAR-operated synapses in GCs from epileptic rats is triggered by the arrival of mossy fiber axons as shown previously for CA3 pyramidal cells during development (Marchal and Mulle, 2004). This reinforces and extends the unique relationship between mossy fibers and KAR-operated synapses (Monaghan and Cotman, 1982; Represa et al., 1987; Castillo et al., 1997; Vignes and Collingridge, 1997; Mulle et al., 1998; Cossart et al., 2002; Marchal and Mulle, 2004). Postsynaptic KARs, unlike the widespread expression of postsynaptic AMPARs, are expressed in specific cell types (Huettner, 2003; Lerma, 2003). Among principal cells of the control hippocampus, only CA3 pyramidal cells use KARs as postsynaptic mediators of ionotropic glutamatergic transmission (Lerma, 2003). Additionally, our results show that mossy fiber sprouting induces a reorganization of KAR-mediated synaptic transmission in the hippocampus of chronic epileptic rats.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Desensitization of kainate receptors reduces synchronized burst discharges in granule cells from epileptic rats. A, Drawing of a hippocampal slice depicting the position of the extracellular recording pipette (granule cell layer) and the stimulating electrode (arrows, lucidum of CA3b to antidromically stimulate granule cells). Experiments were performed in 6 mm K⁺ containing ACSF and 10 μm bicuculline. B, Digitally averaged traces showing that antidromic stimulations of mossy fibers evoke a single population spike (61 averaged traces, left) in a control slice, and a population spike followed by an interictal burst discharge (67 averaged traces; middle) in a slice from an epileptic rat. Note that bath application of 10 μm SYM 2081 (which desensitizes KARs) reduces the severity of the burst discharge in the epileptic slice (as shown by 67 averaged traces; right).

Properties of aberrant EPSC_KA

The EPSC_KAs generated in GCs in chronic epileptic rats share common physiological features with EPSC_KA described in CA3 pyramidal cells and CA1 interneurons in the control hippocampus (Cossart et al., 2002). KARs in GCs from epileptic rats can be activated by the quantal release of glutamate, because we were able to pharmacologically isolate miniature EPSC_KA and to minimally evoke these events. These unitary events display a slow time course and a small amplitude when compared with EPSC_AMPA (Castillo et al., 1997; Vignes and Collingridge, 1997; Frerking et al., 1998; Cossart et al., 2002). Their slow time course is most likely attributable to intrinsic features of KARs as observed in other systems (Lerma, 2003) and not to changes of intrinsic membrane properties of GCs after epilepsy. Accordingly, the kinetics of EPSC_AMPA was not significantly different in epileptic and control GCs. Furthermore, we show that focal GC stimulations or minimal stimulations of recurrent mossy fibers evoked either slow KAR- or fast AMPAR-mediated EPSCs but not mixed AMPAR/KAR-mediated EPSCs. These observations suggest that, in epileptic GCs, AMPARs and KARs are likely to be clustered into different mossy fiber synapses in contrast to control CA3 pyramidal cells in which most KARs are colocalized with AMPARs (Cossart et al., 2002).

Implications for temporal lobe epilepsy

Previous studies emphasized the role of KARs in epileptiform activities induced by acute administration of convulsing agents in vitro or in vivo (Ben-Ari, 1985; Gaiarsa et al., 1994; Mulle et al., 1998; Ben-Ari and Cossart, 2000; Smolders et al., 2002; Rogawski et al., 2003) and in particular the effects of kainate administration that trigger a seizure and brain damage syndrome that mimics human TLE (BenAri, 1985). In the present study, we were concerned by the fate of KAR-mediated synaptic transmission in chronic epileptic rats that developed spontaneous behavioral seizures. In animal models and in patients with TLE, mossy fiber sprouting leads to the formation of a functional recurrent excitatory circuit between granule cells (Tauck and Nadler, 1985; Wuarin and Dudek, 1996; Molnar and Nadler, 1999; Buckmaster et al., 2002; Scharfman et al., 2003) that support epileptiform bursts (Patrylo and Dudek, 1998; Hardison et al., 2000; Gabriel et al., 2004). Mossy fiber sprouting occurs as a compensatory mechanism to the death of hilar mossy cells and the resulting loss of glutamatergic inputs in the inner molecular layer of the dentate gyrus. Our results show that the overall frequency of mEPSCs in GCs is not significantly different between control and chronic epileptic animals, suggesting that mossy fiber sprouting quantitatively compensates for the loss of glutamatergic excitatory inputs. Despite this quantitative compensation, we found a profound change between GCs from control and chronic epileptic rats, because only the latter use KARs as postsynaptic mediators of ionotropic glutamatergic synaptic transmission. What could be the functional consequences of this change? Because EPSC_KA from recurrent mossy fibers represent half of the glutamatergic currents, activation of KARs in addition to AMPARs (Okazaki et al., 2001) should contribute to the epileptiform bursts supported by the recurrent mossy fiber network. Although additional studies are required to substantiate this, the inhibitory effect of a selective KAR blocker on synchronized activities generated by the recurrent mossy fiber network is in keeping with this hypothesis. These observations, together with previous studies, are compatible with an anti-epileptic strategy based on kainate receptor antagonists. Furthermore, the present observations show that the functional consequences of reactive plasticity on glutamatergic synaptic transmission should not only be evaluated in terms of quantitative changes but also qualitative changes. Indeed, whereas control GCs of the dentate gyrus have only fast EPSC_AMPA, GCs from epileptic rats have fast EPSC_AMPA and slower EPSC_KA. Computing studies have shown that EPSC_AMPA and EPSC_KA encode different features of afferent activity (Frerking and Ohligre-Frerking, 2002): EPSC_KAs, with slow kinetics, generate a large tonic depolarization when activated by afferent inputs, whereas EPSC_AMPAs conserve the temporal structure of the signal. Thus, the activation of both AMPARs and KARs at recurrent mossy fiber synapses would greatly modify the computational properties of these cells.

In conclusion, we show that, in addition to axonal rewiring, mossy fiber sprouting induces a reorganization of KAR-mediated synaptic transmission in the hippocampus from chronic epileptic rats together with a shift in the nature of glutamatergic transmission in GCs. This, together with other alterations that occur in the hippocampus (Nadler, 2003), might be important for the pathogenesis of temporal lobe epilepsy.