Single Bursts of Individual Granule Cells Functionally Rearrange Feedforward Inhibition

Paired recordings to test the effects of physiologically relevant burst activity in single GCs on identified postsynaptic CA3 cells

We tested the postburst consequences of single MF bursts in various types of postsynaptic CA3 neurons using unitary responses in paired patch-clamp recordings (for the experimental arrangement, see Fig. 1A, left). For the presynaptic side, we obtained precise control of the synaptic outputs of single presynaptic MFs by using two different approaches. First, direct intracellular or cell-attached recordings from MF terminals allowed us to assess the properties of unitary EPSCs from individual DG GCs to CA3 neurons (Szabadics and Soltesz, 2009). Second, somatic recordings from ectopic GCs (CA3 GCs), whose synaptic properties are indistinguishable from those of DG GCs (Szabadics et al., 2010), allowed for a reliable comparison of MF-to-CA3 neuron EPSCs across multiple protocols, which require long-term recordings. The results from the two types of presynaptic MF recordings were comparable and were therefore grouped together for analysis (for direct comparison of the effects of the two MF sources, see below). The postburst potentiation was primarily assessed by comparing the relative amplitudes of preburst control and postburst test MF-EPSCs (e.g.; Fig. 1B, y axis) with the test responses following the bursts within hundreds of milliseconds to several seconds (x axis in Fig. 1B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Single, brief MF bursts potentiate monosynaptic MF-EPSCs in CA3 FF-INs for several seconds. A, Schematic representation of the recording configuration for unitary MF responses in CA3 neurons. Presynaptic patch-clamp recording was obtained from a single giant MF terminal (most of these originate from DG GCs; Fig. 1-1) or a single CA3 GC (somatic recording; these cells are rarer but easier to record than the MF terminals); the postsynaptic cell was an FF-IN. MF output was first measured with a control stimulus, followed by a single 2–20 AP burst (150 Hz); after various time delays (0.1–13.5 s), which approximated the typical physiological inactivity periods in GCs in vivo, a test response was evoked. Right, Example traces of single-AP-evoked control and test EPSCs in an identified FF-IN (the illustrated recordings were from an IvyC; for the anatomy of the presynaptic and postsynaptic cells, see Fig. 1-1). In most cases, control and test pulses contained 3 APs at 20 Hz to gain insight into possible changes in short-term plasticity after the bursts; plots in this and subsequent figures show responses to the first APs in the triplet, except in Figure 4C, D. B, Time course of the relative amplification of monosynaptic MF-EPSCs after single 15 AP presynaptic bursts in the same individual MF. Relative postburst EPSC amplitudes are shown (i.e., control relative amplitudes are 1, dashed line). The x axis indicates the time of the test pulse after the burst. The graph includes all data points (n = 280, first APs) from identified FF-IN pairs (n = 78 pairs; including IvyCs: n = 55 pairs; AACs: n = 10; PV+BC: n = 5; CCK+IN, n = 8; for separate analysis of the burst-induced amplification in different postsynaptic cell types, see Figs. 1-1, 1-2, 1-3, and 1-4). Insets, Example test responses in blue at different postburst delays; same postsynaptic IvyC as in A; control traces are black. C, Dependence of the postburst potentiation on AP numbers within the burst (from n = 99 data points from n = 28 pairs). Brown curve indicates the exponential fit of the data (R² = 0.966). Gray area represents the typical range of GC bursts in vivo. Insets, Example test and control traces and the time course of the changes of the responses after 3, 5, 9, and 15 AP bursts. D, Effects of single presynaptic bursts consisting of 8 APs at 20, 40, 80, and 150 Hz, on the same MF-FF-IN pairs (n = 7). E, Effects of single bursts, whose patterns were obtained from in vivo recorded identified GCs: #1, 6 APs with 3.7, 7.5, 4.54, 6.98, and 5.88 ms interspike intervals (Henze et al., 2002); #2, 4 APs with 4, 6, and 9 ms interspike intervals (Pernía-Andrade and Jonas, 2014); #3, 5 APs with 6, 8, 11, and 15 ms interspike intervals (Diamantaki et al., 2016); for comparison, the effect of a 6 AP 150 Hz burst is also shown. The effects were measured with each realistic burst protocol in the same pairs (n = 8 pairs, 3.6 s after burst).

In total, we tested the effects of single bursts (in our standard protocol with 15 AP at 150 Hz) in 78 connected pairs in which the postsynaptic interneurons were identified as belonging to the previously reported feedforward category (FF-IN), including ivy cells (IvyCs, n = 55 pairs), axo-axonic cells (AACs, n = 10 pairs), PV-expressing, fast-spiking basket cells (PV+BCs, n = 5 pairs), and regular-spiking CCK-expressing interneurons (CCK+INs, n = 8 pairs, including n = 3 unequivocally identified CCK basket cells). The differences in the numbers of tested pairs reflect neither the occurrence of the cell types nor the probability of connectivity from GCs (Szabadics and Soltesz, 2009). The basal properties of MF-EPSCs onto FF-INs and their short-term plasticity, ranging between slight facilitation and slight depression (see below), were consistent with previous results (Tóth and McBain, 1998; Tóth et al., 2000; Szabadics and Soltesz, 2009; Torborg et al., 2010). The postsynaptic cells were recorded in voltage-clamp mode, minimizing confounds from spiking activity of the recorded postsynaptic cells to the measured changes.

Seconds-long potentiation of MF-EPSCs in FF-INs after single MF bursts

First, we studied the postburst consequences of single MF bursts consisting of 15 APs in 100 ms (150 Hz). Such presynaptic AP bursts had surprisingly large effects on the MF-EPSCs in FF-INs, resulting in a tripling of the amplitudes of the test compared with the control responses for several seconds after the burst (Fig. 1A,B; Figs. 1-1, 1-2, 1-3, and 1-4). Specifically, the effect of the bursts in each pair (n = 78 pairs) was examined at 3 or 4 different postburst times for the test responses, and the average increase in test over control responses was 3.09 ± 0.13-fold between 1.5 and 6.7 s after the burst (control response: −43.8 ± 4.6 pA; test: −109 ± 8.6 pA; p = 5 × 10⁻³⁶, t₍₁₅₃₎ = 16.58, paired t test, total number of postburst test responses at various postburst times, n = 154; Fig. 1-1).

As the plot in Figure 1B indicates, the burst-induced increase in the MF-EPSCs in FF-INs was not instantaneous, and it gradually developed during the first second after presynaptic burst. After this initial increase, the burst-induced MF-EPSCs remained at a similarly enhanced level for about 8 s (comparison of the responses between 0.8 and 8 s after the burst, in 6 bins by one-way ANOVA: p = 0.54, F_(5,198,203) = 0.814), before the MF-EPSC amplitudes returned to baseline, control values (p = 0.081, t₍₅₎ = 2.184, n = 6 timings, t test, 9.5–13.5 s after burst). It should be noted that the postburst potentiation shown in Figure 1B is distinct from the classical post-tetanic potentiation (Alle et al., 2001; Mori et al., 2007) not only because the latter typically requires considerably more intense presynaptic activity than our physiologically inspired burst protocol, but also because of its temporal profile since post-tetanic potentiation is largest immediately after the high-frequency stimulation and then continuously decays back to baseline. Interestingly, despite the specialized roles and synaptic-cellular mechanisms of the various interneurons within hippocampal circuits (Somogyi et al., 2014), the MF bursts had similar effects on all the postsynaptic FF-IN groups that we examined, including IvyCs, AACs, CCK+INs, and PV+BCs (relative increase in test/control responses: 3.15 ± 0.16, 2.88 ± 0.32, 2.98 ± 0.64, and 2.99 ± 0.3-fold amplification, respectively, one-way ANOVA, p = 0.91, F_(3,150,153) = 0.187; Figs. 1-1, 1-2, 1-3, 1-4).

Robust effects of single bursts consisting of only a few (3–5) APs

Although the results above were obtained using bursts in which the firing frequency represented typical GC burst (Henze et al., 2002; Pernía-Andrade and Jonas, 2014; GoodSmith et al., 2017) and these bursts contained considerably fewer APs than those in previously used paradigms (Alle et al., 2001; Mori et al., 2007), the 15 APs per burst that we used in the experiments described above were still more than what is found in typical GC bursts in vivo (2–7 APs) (Henze et al., 2002; Pernía-Andrade and Jonas, 2014). Therefore, we tested the effectiveness of shorter, truly physiological single bursts, aiming to determine the minimal presynaptic activity range that still results in postburst potentiation of MF-EPSCs. In these experiments, the postburst timing of the test MF-EPSC was kept constant (3 s) and the AP number within the burst was systematically varied between 1 and 20 spikes (e.g., effects of 2-5-9 or 1-4-7-10 or 1-9-17, etc.; AP bursts were tested in any given cell pair, for a total of n = 99 combinations). As shown in Figure 1C (n = 28 pairs), even as low as 2–5 APs resulted in measurable postburst potentiation. Single exponential fits to the measured potentiation predicted that the amplification would reach 50% of its maximal value with a theoretical 2.92 AP burst (±5% confidence range: 2.29–3.7 APs; fit R² = 0.966, p = 3.8 × 10⁻¹⁰). A single burst with 5 APs activated 68.9 ± 8% of the maximal potentiation, and 10 AP bursts increased this value to 90.1 ± 10.2%. Control pulses alone (i.e., omitting the bursts from the protocol) did not evoke potentiation (1.05 ± 0.12; Fig. 1C, 0 AP burst point). These data demonstrate that physiological GC AP bursts are well within the dynamic range for the induction of this form of MF plasticity in FF-INs. Interestingly, the effects of shorter bursts also lasted for several seconds (Fig. 1C, inset).

Next, to explore the frequency-dependency of the single burst-induced potentiation, we evoked 8 presynaptic APs at 20, 40, 80, and 160 Hz in the same FF-IN-pairs (n = 8 pairs; Fig. 1D). These presynaptic stimulation protocols did not include preburst control pulses, and the postburst amplitudes were compared with the first responses during the bursts. Interestingly, the amplification showed only a moderate frequency dependence (linear fit, R² = 0.769, p = 0.08).

Finally, we stimulated presynaptic GCs using bursts that had been observed in in vivo recordings previously by other laboratories (Henze et al., 2002; Pernía-Andrade and Jonas, 2014; Diamantaki et al., 2016). These bursts included 4–6 APs with average firing from 100 to 175 Hz. The timing of the spikes within the burst either precisely replicated the in vivo data (pattern 1) (Henze et al., 2002) or, if the exact timing of each AP within the actually observed bursts was not available, the bursts of spikes were patterned to reflect the previously reported average frequencies and jitter by considering the approximate frequency-adaptation during bursts: pattern 2 (Pernía-Andrade and Jonas, 2014) and pattern 3 (Diamantaki et al., 2016) (Fig. 1E, inset). These in vivo bursts and the burst-protocol that we used above (6 APs at 150 Hz) evoked similar potentiation in the same FF-IN-pairs (Fig. 1E; n = 8 pairs, 3 s after burst, p = 0.82, F_(3,21) = 0.306, one-way repeated measure ANOVA, sphericity assumed).

For the subsequent experiments described below, unless specifically stated otherwise, the more robust 15 AP presynaptic bursts were used to enable us to investigate various aspects and mechanisms of postburst potentiation in FF-INs.

Postsynaptic cell type specificity of the postburst MF-EPSC potentiation

Next, we tested whether MF inputs to other postsynaptic cell types in the CA3 network, including pyramidal cells and non-FF-IN GABA cells, also exhibit strong potentiation after a single short MF burst. In CA3 pyramidal cells, the postburst test responses were increased compared with the preburst controls (1.72 ± 0.26-fold increase, n = 12 pairs, total number of postburst test responses in all pairs at various postburst times: n = 25; p = 0.01, t₍₂₄₎ = 2.798; Fig. 2A,B), but the potentiation was less robust than the tripling of MF-EPSCS observed in FF-INs (compare Fig. 2B with Fig. 1B; the data in Fig. 2B were obtained using either 6 or 15 AP bursts because, in 4 of the tested pyramidal cell pairs, the 15 AP presynaptic MF bursts resulted in spiking of the postsynaptic pyramidal cell even in voltage clamp; control experiments showed that the postburst potentiation in pyramidal cells remained smaller than in interneurons even after 20 AP bursts; Fig. 2A). The difference in the nature of the postburst potentiation between the two cell groups was especially striking when the increase in the test response was expressed with respect to the magnitude of the short-term facilitation of the MF-EPSCs taking place during the bursts. As shown in Figure 2C, D, when normalized to the maximum responses during the burst, the postburst test responses were significantly larger in the FF-INs compared with the pyramidal cells. These results suggest that the postsynaptic responses in CA3 pyramidal cells are more sensitive to the bursts, while they are relatively more affected during the postburst period in FF-INs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Postsynaptic cell type specificity of postburst potentiation of MF-EPSCs. A, Single presynaptic MF bursts, consisting of various numbers of APs, resulted in small postburst potentiation effects on the EPSCs of a representative MF-pyramidal cell pair. Traces represent presynaptic MF terminal APs in bouton-attached recordings (gray) and the postsynaptic responses before and after single burst. Bursts and postsynaptic responses (including preburst control and postburst test EPSCs, and during-burst responses) are shown next to the corresponding traces. B, Time course of the effects of single, brief presynaptic bursts (6 or 15 AP, 150 Hz) on subsequent single-AP-evoked MF-EPSCs in postsynaptic CA3 pyramidal cells (n = 46 data points from n = 12 pairs; compare with Fig. 1B). Inset, Example control and test MF responses at different postburst delays. Anatomy of the postsynaptic pyramidal cell and presynaptic DG GC (back-labeled via the MF terminal recording) is shown in Figure 2-1. C, Summary graph showing the amplitude of the postburst test responses in pyramidal cells and FF-INs relative to the compound maximal amplitudes reached during the bursts (dashed line) in the same synaptic inputs (red represents pyramidal cells, n = 46 data points from n = 12 pairs; blue represents FF-INs with 15 AP bursts; light blue represents FF-INs with short, 3 or 5 AP bursts, n = 10 pairs; the same responses were reanalyzed as in Fig. 1B,C). The relative amplitudes of the preburst control responses (“before”) are shown separately. D, Comparison of MF-EPSCs before, during, and after single presynaptic MF bursts in a representative FF-IN (blue traces) and pyramidal cell (red). Green dashed lines indicate maximum amplitude of the compound EPSCs during the burst for comparison. Gray areas represent the integral areas (charge transfer). Bottom, Graphs represent the activity- and postsynaptic cell type-dependent differences of the time course of charge transfer.

Because the postburst MF-EPSC potentiation was not different between axo-axonic, PV and CCK basket, and ivy cells, we next examined whether the similarity of the effects of single MF bursts extended to non-FF-IN interneurons. The SLCs have negligible local, within-CA3 axons and thus do not participate in feedback or feedforward inhibition but comprise a part of the hippocampo-septal GABAergic projection (Gulyás et al., 1992; Spruston et al., 1997; Jinno et al., 2007). Interestingly, despite the fact that SLCs receive excitatory inputs almost exclusively from MFs (Wittner et al., 2006), the MF synapses on SLCs have extremely low initial release probability, which increases only after sustained presynaptic activity (Szabadics and Soltesz, 2009; Szabadics et al., 2010). Our paired recordings showed that MF synaptic release onto SLCs was increased shortly after the 15 AP MF bursts (Fig. 3A,D). However, in contrast to FF-INs, the amplitudes of the postburst test MF-EPSCs remained small compared with the compound EPSC amplitudes evoked during the bursts and quickly returned to control baseline (Fig. 3). Thus, the bursts were not able to induce the sustained, seconds-long potentiation of MF-EPSCs in SLCs. Importantly, similarly small burst effects were observed, even when the unusually low control synaptic release probability of the MF-SLC connections (Fig. 3B) was artificially increased (Fig. 3C,D), suggesting that the postsynaptic cell type specificity of the amplification was not simply a consequence of the different initial release probabilities of the MF inputs onto the SLCs and FF-INs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Effects of single MF bursts on the unitary MF responses evoked in SLCs, which are GABAergic but are not considered to be FF-INs. A, Control and postburst test MF-EPSCs from an example CA3 GC-SLC pair at three different time points after single presynaptic bursts. Gray bars represent APs, often weak responses among the frequent spontaneous events. B, The MF responses of two SLCs during the burst (15 APS at 150 Hz) in standard solutions (top) and in the presence of presynaptic intracellular CsCl (by replacing 40 mm KCl) and 3.5 mm extracellular Ca²⁺ level (bottom, **). C, A representative experiment, in which reliable synaptic transmission was obtained in a CA3 GC-to-SLC connection by the above modifications in the recording conditions. D, Summary graph showing the amplitude of the postburst responses in SLCs relative to the compound maximal amplitudes during the bursts. Dark blue represents SLC with standard presynaptic recording (n = 61 data points from n = 20 pairs). Gray represents SLC with artificially elevated MF release (n = 15 data points from n = 5 pairs). Blue curve indicates data for FF-INs replotted from Figure 2C to highlight the robust differences between the two GABAergic cell groups. The relative amplitudes of the control responses before the bursts are shown separately. For the anatomy of the presynaptic and postsynaptic cells, see Figure 3-1.

Presynaptic origin of the burst-induced sustained amplification in FF-INs

To better understand the underlying synaptic mechanisms of the single-burst-induced amplification, we tested whether the amplification of the MF-responses in FF-INs was initiated by presynaptic or postsynaptic changes. The probability of failures decreased after single bursts (from 53.3 ± 3% to 25.0 ± 3%; p = 3.5 × 10⁻²⁰, t₍₆₇₎ = 13.12, n = 68 FF-IN pairs, paired t test; Fig. 4A), consistent with presynaptic changes. The rise times and half-widths of the postburst test MF-EPSCs remained unchanged compared with preburst control (625 ± 38 μs vs 583 ± 31 μs, p = 0.25, t₍₅₀₎ = 1.155; 2.46 ± 0.12 ms vs 2.61 ± 0.09 ms, p = 0.13, t₍₅₁₎ = −1.556, paired t test; Fig. 4B). Comparison of the short-term plasticity of the control and test responses evoked using 3 MF APs at 20 Hz before and after the burst was also consistent with a presynaptic locus for the postburst MF-EPSC potentiation. Specifically, the PPR decreased after the burst and remained small (i.e., more pronounced depression) for several seconds (Fig. 4C), similar to the time course of the burst-induced MF-EPSC potentiation in Figure 1B.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Presynaptic origin of the burst-induced potentiation in FF-INs. A, The probability of synaptic response failures decreases after single presynaptic MF bursts. Representative traces of a CA3 GC-CCK⁺ basket cell connection before (black) and after (blue) single bursts (10 traces from each condition). Right, Connected gray symbols represent the failure rates in individual pairs with identified FF-INs (n = 68 pairs) during control and test MF stimuli (1.5–6.7 s postburst delays). Black represents the average data from all FF-IN connections. The failure rates were analyzed from at least 10 trials for each condition. B, Actual and normalized average EPSCs evoked in FF-INs (n = 51 pairs) by single MF APs before and 1.5–6.7 s after single presynaptic bursts. The events were aligned to the presynaptic AP peak. Graphs represent the half-width and 10%–90% rise times of the MF responses in individual FF-INs (gray) and their average (black), which were similar before and after the bursts. C, Example traces demonstrating changes in short-term plasticity of the MF-EPSCs in an IvyC evoked by 3 APs before and 6 s after bursts (control PPR: 1.26; test PPR: 0.45). Summary plot shows data from all in identified postsynaptic FF-INs; the PPR (calculated as the ratio of the average of the second and third amplitudes to the first amplitude) was low during the time when the EPSCs were potentiated. Summary bar graphs represent PPRs of all FF-IN connections before (control) and 1.5–6.7 s after the burst (test). D, Example traces: from IvyC MF pairs; note the similarity of test PPRs (at 6 s postburst) despite the different control PPRs. Right panels: Top, Correlation between the control (initial) PPR and the magnitude of potentiation (linear fit, maroon line; note the logarithmic scaling of the PPR axes). Bottom, Independence of postburst test PPR of the control PPR. Each circle represents individual test responses from identified FF-IN pairs. Circles with gray fillings indicate the representative pairs illustrated in C and D. * marks significant difference between control and test responses (see text for details).

The PPR of the control (initial, preburst) MF responses varied over a relatively wide range across FF-IN pairs (even within the same types) under baseline conditions. The control PPR showed a correlation with the magnitude of the potentiation (R² = 0.438, slope: 1.54 ± 0.12, F_(1,196) = 154.8, p < 10⁻²⁰), indicating that, for connections whose release probability was initially low (i.e., high PPR), single bursts led to larger potentiation, whereas the relative potentiation of the more reliable connections was rather small (Fig. 4D). The postburst test PPRs did not correlate with the initial control PPRs of the same connections (R² = 0.035, slope: 0.139 ± 0.048, F_(1,196) = 8.227, p = 0.0046), primarily because the PPRs after the bursts were invariably low (i.e., highly reliable, depressing responses), suggesting that single bursts result in a high-release state in MF-FF-IN synapses regardless of their initial reliability.

Investigations into plasticity mechanisms

To characterize postburst plasticity in FF-INs further, we first explored whether the postburst potentiation at unitary MF-FF-IN synapses was present at a lower extracellular Ca²⁺ concentration (1.3 mm) that may be more physiological (Rancz et al., 2007; Lorteije et al., 2009) than the one used in the standard recording solution (2 mm) in the above-described experiments. In 1.3 mm extracellular Ca²⁺ (Mg²⁺ concentration was kept unchanged at 2 mm), as expected, the control response amplitude decreased and PPR increased (n = 8 pairs, p = 0.012, t₍₇₎ = −3.38 and 0.011, t₍₇₎ = −1.97, respectively, paired t test), but the postburst potentiation of the MF-FF-IN inputs not only persisted but increased (from 1.81 ± 0.16 test/control amplitudes to 3.55 ± 0.48 in the same pairs with normal and low Ca²⁺ conditions, respectively, p = 0.002, t₍₁₂₎ = −3.862, paired t test; Fig. 5A). The apparently larger postburst potentiation in lower extracellular Ca²⁺ was due to a larger decrease of the control EPSCs compared with the postburst EPSCs (−76.4 ± 1.9% vs −54.3 ± 5.8%, p = 0.0017, t₍₁₂₎ = −4.034, paired t test), consistent with a presynaptic origin for the plasticity. Next, we tested the effects of presynaptic Ca²⁺ chelation on postburst plasticity. Presynaptic loading of the Ca²⁺ chelator EGTA (0.5–2.5 mm; the final concentration in the presynaptic terminal was unknown, as the CA3 GC somatic recording site to presynaptic terminal distance varied) (Szabadics et al., 2010) strongly reduced the control response (p = 4 × 10⁻⁴, t_(15.03) = 4.615, paired t test) but did not prevent the single-burst-induced amplification (p = 0.0039, t₍₁₅₎ = 3.409, t test; Fig. 5B). Therefore, postburst potentiation at MF-FF-IN synapses persisted at physiological extracellular Ca²⁺ levels and was consistent with the data shown in Figure 4D. In addition, consistent with the results of the 1.3 mm Ca²⁺ and EGTA experiments, when the extracellular Ca²⁺ concentration was elevated to 5 mm, the baseline responses increased by 178 ± 33%, whereas the effect on postburst EPSPs was smaller at 104 ± 28% (p = 0.012, t₍₁₄₎ = 2.888, paired t test, data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Investigations into mechanisms of postburst plasticity in FF-INs. A, Persistence of the postburst potentiation in lower (1.3 mm; allegedly physiologically more relevant) extracellular Ca²⁺ levels. Example traces are shown from a CA3 GC-IvyC pair in which the effects of single presynaptic bursts were first tested in the presence of standard 2 mm Ca²⁺ concentrations and then in 1.3 mm Ca²⁺ (preburst control responses: black; test responses: blue, 3.6 s after burst). Bar graphs represent average data from all pairs (n = 8); the lower Ca²⁺ concentration decreased the response amplitudes and increased the PPRs, but the potentiation persisted. B, Example traces from a CA3 GC-AAC pair in which the effects of single presynaptic bursts were tested while the presynaptic CA3 GC was recorded with 1 mm intracellular EGTA to chelate Ca²⁺. Bar graphs represent average data from all presynaptic EGTA pairs (n = 8). Presynaptic EGTA strongly decreased the control responses, but the burst-induced potentiation persisted. C, Example traces illustrate the accelerated and more precise release before and after single presynaptic bursts from the same pair; average presynaptic APs and individual EPSCs are shown; mean and variance of the delay are also indicated, with error bars; failures were excluded for clarity. Right panels, Plots of the mean and variance values of the synaptic onset delay in each MF-FF-IN pair before and after single bursts (connected gray symbols) and their average (black). D, The DAG analog phorbol ester PDBu (1 μm), which promotes vesicle priming, increased MF-EPSC amplitude before bursts (note the different scale bars) and prevented burst-induced potentiation in a representative pair. Subsequent reduction of the release probability following the application of decreased extracellular Ca²⁺ (1 mm) in the PDBu-containing perfusing solution significantly reduced the average amplitudes, but burst-induced amplification remained negligible in the same pair. Summary bar graphs show that bursts were no longer effective in eliciting potentiation in the same pairs in the presence of PDBu regardless of the release probability; furthermore, PDBu also prevented the burst-induced decrease in the delay of the responses (bottom right). Symbols represent changes in synaptic delays in control conditions and in the presence of PDBu. Bars represent average data. * marks significant difference.

Next, we investigated whether the delay between the presynaptic spike and the onset of the postsynaptic response in FF-INs was altered by the single bursts of MF APs. Our analysis showed that the delay of the test responses was significantly shorter than the delay of the control responses in MF-FF-IN pairs (control: 0.949 ± 0.023 ms; test: 0.868 ± 0.023 ms, n = 57 responses, p = 9 × 10⁻¹², t₍₅₆₎ = 8.565, paired t test; Fig. 5C, top). Furthermore, the variance of the delays was also significantly decreased (from 0.0294 ± 0.0023 ms² to 0.018 ± 0.0018 ms², p = 3 × 10⁻⁵, t₍₅₆₎ = 4.544; Fig. 5C, bottom; the CV of the delay also changed: from 0.176 ± 0.008 to 0.145 ± 0.008, p = 4.5 × 10⁻⁴, t₍₅₆₎ = 3.729). Therefore, the single, brief burst of MF APs results in accelerated, more precise, and more reliable release.

In subsequent experiments, we tested whether the burst-induced plasticity could be occluded by the phorbol ester PDBu, thought to promote vesicle priming and saturate the capacity of the release machinery (Rhee et al., 2002; Lou et al., 2008; Fioravante et al., 2014; Taschenberger et al., 2016). PDBu (1 μm) increased the control response amplitude before the bursts (from −39.6 ± 5.3 pA to −146.3 ± 52.2 pA, p = 0.011, t₍₁₂₎ = 2.646) and decreased the PPR (from 1.51 ± 0.10 to 0.61 ± 0.04, p = 0.0004, t₍₁₂₎ = 4.462, t test), and there was no significant burst-induced potentiation of the already enhanced MF-EPSCs in FF-INs in the presence of PDBu (test/control amplitude: 1.32 ± 0.15, p = 0.21, t₍₁₂₎ = 2.11, t test; Fig. 5D). Furthermore, single MF bursts failed to potentiate the test responses with respect to the preburst control responses in the presence of PDBu, even in reduced extracellular Ca²⁺ (1–1.6 mm), applied to compensate for the amplitude-enhancing effects of PDBu (test/control amplitude: 1.4 ± 0.15, p = 0.19, t₍₁₅₎ = 2.67; the reduced Ca²⁺ concentration alone would have been expected to increase amplification, see Fig. 5A). In addition, application of PDBu also prevented the bursts from accelerating the synaptic release (onset delay of the control and test responses: before PDBu: 1.005 ± 0.069 ms and 0.944 ± 0.082 ms, p = 0.017, t₍₈₎ = 3.01, paired t test; in PDBu: 0.96 ± 0.07 ms and 0.967 ± 0.084 ms, p = 0.87, t₍₈₎ = 0.165; ANOVA one-way repeated measure, multivariate test: p = 0.048, F₍₆₎ = 4.851, between-subject effects: p = 9.7 × 10⁻⁷, F_(1,8) = 177.01; Fig. 5D, bottom right). Together, these results suggested occlusion, and thus at least a certain degree of mechanistic convergence, between the burst and PDBu effects, perhaps in the form of enhanced vesicle priming at MF synapses on FF-INs.

We also investigated whether mechanisms implicated in other forms of plasticity at MF synapses were involved in the postburst potentiation described in this paper. PKC and Munc13 proteins are known to be involved in regulating release at many synapses (Betz et al., 1998; Augustin et al., 1999; Wu and Wu, 2001; Rosenmund et al., 2002; Junge et al., 2004; Korogod et al., 2007; Wierda et al., 2007; Fioravante et al., 2011; Chu et al., 2014), including MFs (Alle et al., 2001; Pelkey et al., 2005; Hainmüller et al., 2014). In agreement with these results, various direct and indirect inhibitors of PKC activity (calphostin-C, intracellular PKC19-36, GF109203X, and U73122, the latter inhibits PLCs that generate diacyl-glycerol for PKC functions) were able to decrease the control EPSCs at MF-FF-IN synapses. Specifically, acting as positive controls, the drug application reduced the control EPSC amplitudes (calphostin-C, with respect to predrug: 60.5 ± 27.4%; U73122: to 37.5 ± 17.5%) and increased the PPR (by 0.53 ± 0.55 and 0.67 ± 0.37, respectively; Fig. 6A); similarly, in the case of experiments with PKC19-36 fragment and GF109203X (these drugs were preincubated), the EPSC amplitudes were 37.9 ± 3.2% and 48.6 ± 5%, respectively, of the untreated control pairs, and the PPRs were increased by 0.32 ± 0.26 and 0.32 ± 0.11, respectively. In contrast, another type of PKC blocker, Go6976, did not alter the control MF responses (control EPSC amplitude with respect to predrug control: 108.6 ± 55.2%; PPR: 0.24 ± 0.47).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Summary of the pharmacological experiments aimed at probing the potential involvement of various plasticity pathways in the postburst potentiation in FF-INs. A, Calphostin-C (1 μm, n = 12) is a PKC (and Munc13) inhibitor that acts on DAG-binding domains. U73122 (2.5 μm, n = 8) is a phospholipase C inhibitor. Both inhibitors affected baseline release as indicated by the attenuation of the control response amplitudes indicating the involvement of these pathways in the regulation of release at these synapses, serving as positive internal controls indicating drug effectiveness. PKC19-36 (100 μm, n = 7, synthetic autoinhibitory domain, applied intracellularly) and GF109203X (1 μm, n = 11, acts on the ATP binding site; slices were preincubated for at least 60 min) are selective PKC inhibitors. The small control amplitudes and large PPRs provided positive internal controls for these inhibitors. The averages of all control responses are also shown for comparison (because these inhibitors required intracellular application and pretreatment). DCG IV is an mGluR2/3 agonist (1 μm, n = 7) that selectively inhibits MF responses. * marks significant difference in the control (before burst) EPSC amplitudes in control conditions and after drug application. B, Average amplification at 1.5–6.7 s after single 15 AP bursts in the presence of various pharmacological agents. In addition to the above drugs, the burst-induced amplification of MF-EPSCs is shown in the presence of Go6976 (0.25 μm, n = 8, a subtype-selective PKC inhibitor), KT5720 (200 nm, n = 10, a PKA inhibitor that works by competing with ATP binding; PKAs are involved in pathways that regulate presynaptic release, including RIM proteins, whose PKA-dependent phosphorylation promotes vesicle priming and PKAs are also involved in posttetanic potentiation mechanisms at MF synapses onto DG inhibitory cells) (Alle et al., 2001), PKI (2.5 μm intracellularly, PKA inhibitory fragment 6–22 amide, that binds to the substrate site) (Hashimotodani et al., 2017), AMN082 (1 μm, n = 6, selective mGluR7 agonist), MSOP (150 μm, n = 8, selective inhibitor of Group III mGluRs including mGluR7), or BINA (5 μm, n = 12, selective positive allosteric modulator of mGluR2). For additional positive control effects, see Results. Postburst potentiation in the absence of these drugs (“control amplification”) is also shown for comparison. None of these pharmacological modifiers of known plasticity pathway components was able to significantly block the postburst potentiation.

Importantly, if PKC activity were necessary for the burst-induced potentiation, these various PKC inhibitors should have decreased or eliminated the postburst plasticity. However, the postburst potentiation was not affected in the presence of any of these inhibitors (Fig. 6B, group of bars under the label “PKCs”), despite the presence of positive control effects for most of the drugs. Because Munc13 proteins share both active domains (which are targeted by calphostin-C, PDBu, and native diacyl-glycerol sources) and synaptic functions with PKCs (promoting release), we cannot exclude the possibility that Munc13s were involved in the positive control effects, but it is also unlikely that they were necessary for the burst-induced potentiation. Similarly, we were unable to block the postburst potentiation in FF-INs using the PKA inhibitors KT5720 and presynaptically loaded PKI6–22 fragment (Fig. 6B, bars under the label “PKA”; the lack of an effect on the postburst potentiation was in the presence of positive control effects on the control EPSC amplitudes: KT5720, 51.2 ± 17.8%; and PPR, 0.24 ± 0.25). In addition, neither activation nor inhibition of mGluR7 and mGluR2/3 occluded or eliminated the postburst potentiation (Fig. 6B, bars labeled “mGluRs”) (AMN082, control effect on EPSC and PPR, respectively: 70.2 ± 33.7%, 0.51 ± 039; MSOP, 62 ± 21.7%, −0.18 ± 0.17; BINA, 78.6 ± 9.8%, −0.09 ± 0.12; and DCG-IV, 11.3 ± 4%, 0.41 ± 0.38; mGluR7 is specifically expressed at MF synapses on GABAergic cells) (Shigemoto et al., 1997; Pelkey et al., 2008). The fact that none of these pharmacological manipulations of PKC, PKA, and mGluRs was able to significantly modify postburst potentiation (note the remarkably stable level of potentiation in the presence of various drugs in Fig. 6B) suggested that the single, brief burst-induced MF plasticity at FF-IN synapses described in this study may involve a potentially noncanonical molecular pathway that will need to be identified in the future.

Seconds-long potentiation of FF-IN activity in the CA3 network after individual MF bursts

Our results suggest that FF-INs and pyramidal cells in the CA3 area respond to single MF bursts differently, with weak (strong) short-term facilitation of the unitary EPSCs but strong (weak) postburst potentiation in FF-INs (pyramidal cells) (Fig. 2B–D). In other words, the seconds-long MF-EPSC plasticity in FF-INs appear to reflect the recent passage of a MF burst, whereas changes in MF-EPSCs in pyramidal cells preferentially take place during the MF bursts themselves. To test these ideas further, in the final series of experiments, we examined whether a single brief burst of APs in an individual MF is able to increase FF-IN activity in the network by measuring the incidence of diIPSCs in pyramidal cells that were not themselves synaptically connected to the activated MF (for synaptic delays and kinetics of diIPSCs after APs in the MF, see Fig. 7E) (Neubrandt et al., 2017). The probability of diIPSCs was modestly but significantly increased in randomly selected pyramidal cells during the bursts of APs in individual MFs (Fig. 7A,B, bar labeled “burst”; diIPSCs after APs in a single MF: control responses: 11.9 ± 3.4%, during bursts: 18.1 ± 1.9%, p = 0.0029, t₍₂₃₎ = −3.329, paired t test, n = 24 pairs of MF and randomly chosen pyramidal cell). In contrast, in the same pairs, the probability of diIPSCs almost tripled for several seconds following the AP burst in the individual MF, with a time course that showed remarkable similarity to the potentiation of unitary MF-EPSCs after bursts (compare Fig. 7B with Fig. 1B; probability of diIPSCs in response to test pulses 1.5–6.7 s after the bursts: 29.4 ± 3.1%, p = 1 × 10⁻⁹, t₍₃₉₎ = −7.944, compared with control preburst responses and p = 2 × 10⁻⁴, t₍₃₉₎ = −4.09, compared with diIPSCs during the bursts, paired t test). These data indicated that a single burst in an individual MF enhanced the probability of IPSCs for several seconds in a population of pyramidal cells that were not directly connected to the activated MFs, presumably through the involvement of one or more unrecorded FF-IN(s). In addition, these results were similar regardless of whether we evoked bursts in a giant MF terminal or in somatically recorded CA3 GCs (Szabadics et al., 2010) (Fig. 7A, bottom right inset, bar graphs; because DG GCs are much more numerous than CA3GCs, the MF terminal recordings must have overwhelmingly come from MFs of DG origin; therefore, these results also showed that the burst-related plasticity in FF-INs was similar with CA3 GC or DG GC inputs). In contrast, when we assessed the effect of single bursts in individual MFs on pyramidal cell activation by recording disynaptic EPSCs from CA3 FF-INs, we found a strong increase in the probability of diEPSCs during the bursts (control: 12.6 ± 5.5%, during burst: 26 ± 8.1%, n = 6 pairs, p = 0.015, t₍₅₎ = −3.649, paired t test), but no enhancement of the probability of diEPSCs after the bursts (Fig. 7C,D; probability of diIPSCs after test pulses 1.5–6.7 s following the burst: 10.6 ± 2.9%, p = 0.61, t₍₁₁₎ = 0.521, paired t test). These results supported our hypothesis that pyramidal cells were more sensitive to the bursts, whereas the inhibitory network preferentially represented the recent past of MF burst activity.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Differential effects of single bursts in individual MFs on diIPSCs and diEPSCs. A, Example traces: Synaptic events in a single randomly chosen CA3 pyramidal cell during the stimulation of a simultaneously recorded single CA3 GC that was not monosynaptically connected to the pyramidal cell; the majority of the outward events are disynaptic GABAergic responses (diIPSCs; at −50 mV using low chloride internal solution in postsynaptic cells), which presumably originate from one or more nonrecorded FF-IN(s); there is increased probability of diIPSC after single bursts in the MF (test). Black average traces represent the increased weight of disynaptic inhibition after the bursts. Right, Bar graphs represent summary data from diIPSC experiments. The bar graphs illustrate two types of diIPSC pairs: one in which the presynaptic MF source was a MF terminal (presumably originating from DG GCs), the other in which the MF source was a somatically recorded CA3 GC. There is a lack of differences in the burst-induced changes in the diIPSCs, suggesting that these two MF recording configurations induce similar burst-dependent effects on the CA3 feedforward inhibitory circuit. B, Summary data indicating that the postburst time course of the increase in the probability of single MF-evoked diIPSCs. There is similarity of the time course to that of the burst-induced potentiation of the MF-EPSCs in FF-INs in Figure 1B. The increase in the probability of the diIPSCs during the burst (dark blue) was smaller than it was after the burst, consistent with the FF-IN data in Figure 2C. C, Example traces: diEPSC events recorded from a nonidentified fast-spiking interneuron during the activation of a CA3 GC; these diEPSCs presumably reflect the spiking of one or more nonrecorded pyramidal cell(s); there are frequent diEPSCs during the high-frequency bursts (inset), whereas the probability of diEPSCs remained unchanged after the bursts. D, Summary data: burst effects on the probability of diEPSCs evoked by MFs before, during (bars), and after bursts in the same individual MF (average data of 6 pairs, with 15 AP bursts at 150 Hz); note the difference compared with the data in B, and the similarity to the MF-EPSC pyramidal cell data in Figure 2C. * mark significant difference in the probability of disynaptic events relative to control probabilities (before burst). E, Distribution of the onset time delays of individual postsynaptic events measured from the time of the presynaptic AP, including monosynaptic EPSCs in postsynaptic FF-INs (gray), diIPSCs in CA3 pyramidal cells (blue), and diEPSCs in GABAergic cells (red). The 10%–90% rise time and decay time constants of the diIPSCs and diEPSCs are shown below. Each data point represents an individual pair.