At immature mossy-fiber-CA3 synapses, correlated presynaptic and postsynaptic activity persistently enhances GABA release and network excitability via BDNF and cAMP-dependent PKA

Abstract

In the adult rat hippocampus, the axons of granule cells in the dentate gyrus, the mossy fibers (MF), form excitatory glutamatergic synapses with CA3 principal cells. In neonates, MF release into their targets mainly GABA, which at this developmental stage is depolarizing. Here we tested the hypothesis that, at immature MF-CA3 synapses, correlated presynaptic [single fiber-evoked GABA(A)-mediated postsynaptic potentials (GPSPs)] and postsynaptic activity (back propagating action potentials) may exert a critical control on synaptic efficacy. This form of plasticity, called spike-timing-dependent plasticity (STDP), is a Hebbian type form of learning extensively studied at the level of glutamatergic synapses. Depending on the relative timing, pairing postsynaptic spiking and single MF-GPSPs induced bidirectional changes in synaptic efficacy. In case of positive pairing, spike-timing-dependent-long-term potentiation (STD-LTP) was associated with a persistent increase in GPSP slope and in the probability of cell firing. The transduction pathway involved a rise of calcium in the postsynaptic cell and the combined activity of cAMP-dependent PKA (protein kinase A) and brain-derived neurotrophic factor (BDNF). Retrograde signaling via BDNF and presynaptic TrkB receptors led to a persistent increase in GABA release. In "presynaptically" silent neurons, the enhanced probability of GABA release induced by the pairing protocol, unsilenced these synapses. Shifting E(GABA) from the depolarizing to the hyperpolarizing direction with bumetanide failed to modify synaptic strength. Thus, STD-LTP of GPSPs provides a reliable way to convey information from granule cells to the CA3 associative network at a time when glutamatergic synapses are still poorly developed.

Figure 1.

GABA released from MF terminals depolarizes principal cells. A, Synaptic responses (successes plus failures) evoked at three different holding potentials (to the left of the traces) by stimulation of granule cells in the dentate gyrus in control (○) and in the presence of bumetanide (·, 10 μm). B, Plot of GPSP amplitudes as a function of membrane potentials in control (○, n = 10) and in the presence of bumetanide (·, n = 9). Note that in bumetanide, E_GPSP shifted toward more negative values (−71.5 mV vs −47.6 mV). C, GPSPs obtained in control and in the presence of bumetanide were sensitive to l-AP4 (10 μm) and to picrotoxin (100 μm). In bumetanide, GPSPs were recorded at −30 mV to increase the driving force for Cl⁻. D, E, Each symbol represents E_GPSPs (D) and V_R (E) of individual cells obtained in control (○) and in the presence of bumetanide (·). The average values are shown on the left of each group. *p < 0.05; **p < 0.01; ***p < 0.001.

Figure 2.

Spike-timing dependent LTP induced by pairing MF GPSPs with postsynaptic spiking. A, Schematic representation of the experimental design. B, Spike-timing protocol. GPSP preceded the postsynaptic spike by 15 ms (Δt). C, D, Peak amplitude of MF GPSCs in presynaptically silent (C) and low probability neurons (D) evoked before and after pairing (arrows at time 0) as a function of time. Insets above the graphs represent individual (top traces) and averaged GPSCs (bottom traces) evoked before and after pairing. E, Summary plot of mean GPSCs amplitude recorded before and after pairing versus time (n = 12). F–I, Amplitude (F), successes (G), paired-pulse ratio (H), and inverse squared of CV (I) measured in individual cells before (○) and 40 min after pairing (·). Bigger symbols represent averaged values. ***p < 0.001.

Figure 3.

Pairing-induced increase in GPSPs slope was associated with enhanced firing probability. A, Representative traces from one neuron showing the rising phases of GPSPs (evoked at 0.1 Hz) during pairing. Note the progressive increase of GPSPs slope which occasionally gave rise to action potentials (with <15 ms delay). B, The GPSPs slope obtained before and after pairing (arrows at time 0) is plotted as a function of time (n = 7). C, Superimposed traces from a single neuron before and after pairing. D, Each symbol represents the mean value of GPSPs slope obtained from individual cells before (○) and 15 min after pairing (·). Larger symbols represent averaged values (n = 7). E, Representative examples of GPSPs (average of 15 traces) evoked in the presence bumetanide before and after pairing. F, Each symbol represents the mean value of GPSPs slope obtained from individual cells before (○) and 15 min after pairing (·) in the presence of bumetanide. Bigger symbols represent averaged values (n = 7). **p < 0.01.

Figure 4.

Spike-timing dependent LTD induced by pairing MF GPSPs with postsynaptic spiking. A, The inset represents the spike-timing protocol. The postsynaptic spike (post) preceded the GPSP (pre) by 15 ms. Below, Individual (top) and averaged (bottom) traces of GPSCs evoked before (Control) and after pairing. B, In the top graph, the peak amplitudes of GPSCs (shown in A) obtained before and after pairing (arrows at time 0) are plotted against time. In the bottom graph, summary plot of GPSCs amplitude versus time (n = 8). C–F, Amplitude (C), successes (D), paired-pulse ratio (E) and inverse squared of CV (F) measured in individual cells before (○) and after pairing (·). Bigger circles represent average values. (n = 8 for each group except for PPR in which n = 5). *p < 0.05; **p < 0.01.

Figure 5.

Temporal window for spike-timing dependent enhancement and reduction in the efficacy of MF GPSCs. Plot of GPSC amplitude (as percentage of controls) from individual cells (○) as a function of Δt. Closed symbols represent the averaged values for each Δt. No changes in amplitude were detected when the postsynaptic action potential was coincident with presynaptic GPSPs (Δt = 0 ms). Note that the magnitude of synaptic potentiation reached the maximum at Δt = 15 ms which corresponds to the peak of GPSP.

Figure 6.

Pairing-induced potentiation requires a postsynaptic rise of intracellular calcium via voltage-dependent calcium channels. A, Averaged traces of GPSC evoked before and 15 min after pairing in neurons loaded with intracellular BAPTA (20 mm) or exposed to extracellular solutions containing nifedipine (10 μm) or d-AP5 (50 μm). B, Pairing-induced changes in the mean amplitude and mean number of successes expressed as percentage of controls from all cells tested (BAPTA, n = 6), nifedipine (n = 5) and d-AP5 (n = 5). **p < 0.01.

Figure 7.

Pairing-induced synaptic potentiation requires the activation of TrkB receptors by BDNF. A, Summary plot showing the mean amplitude of GPSCs versus time in the absence of BDNF (Control; n = 5). Insets above the graph refer to GPSCs (individual traces above and averaged traces below) recorded during the first (left) and the last (right) 5 min. B, Summary plot showing the mean amplitude of GPSC (n = 7) evoked before, during and after bath application of BDNF (bar). The horizontal dashed line represents the mean amplitude of GPSCs recorded in control before BDNF application. Insets above the graph represent superimposed individual (top) and averaged (bottom) traces of GPSCs evoked before (Control) and 15 min after the application of BDNF (40 ng/ml). C–E, Amplitude (C), successes (D) and paired-pulse ratio (E) measured in individual cells before (○) and 15 min after BDNF (·). Larger symbols represent averaged values. F–J, As in Figure 2 but in the presence of K252a (n = 7). *p < 0.05; **p < 0.01.

Figure 8.

Forskolin mimics the effects of BDNF. A, Summary plot showing the mean amplitude of GPSC (n = 6) evoked before, during and after bath application of forskolin (50 μm, bar). Insets above the graph show averaged traces (15 GPSCs from a representative neuron) evoked before and during bath application of forskolin. B–E, Amplitude (B), successes (C), paired-pulse ratio (D) and inverse squared of CV (E) measured before (○) and during bath application of forskolin (·). Larger symbols represent averaged values. *p < 0.05; **p < 0.01.

Figure 9.

Pairing-induced synaptic potentiation requires the activation of cAMP-dependent PKA. Summary plot showing the mean amplitude of GPSC (n = 5; average of 12 traces) evoked in the presence of Rp-cAMPS before and after pairing (arrows at time 0) as a function of time. Insets above the graph represent individual traces of GPSCs evoked before and after pairing.

Figure 10.

Pairing-induced potentiation of GPSCs was prevented by inhibiting PKA in the postsynaptic neuron. A, Individual traces (each is the average of 10 responses) obtained from a CA3 pyramidal neuron recorded with a pipette containing PKI 6–22 (1 μm) before (Control) and 15 min after pairing. B, Summary plot showing the mean amplitude of GPSCs (recorded with a patch pipette containing PKI 6–22) evoked before and after pairing (arrows at time 0) as a function of time. In the presence of PKI 6–22 into the patch pipette, pairing was unable to modify MF-mediated synaptic responses.