A form of long-lasting, learning-related synaptic plasticity in the hippocampus induced by heterosynaptic low-frequency pairing

Abstract

The late, transcription- and translation-dependent phase of long-term synaptic potentiation (L-LTP) at the Schaffer collateral synapse of the hippocampus is an experimental model of the synaptic plasticity underlying long-lasting memory formation. L-LTP is typically induced by homosynaptic tetanic stimulation; but associative forms of learning are likely to require the heterosynaptic pairing of stimuli. Here we describe L-LTP elicited by such heterosynaptic pairing at the Schaffer collateral synapse in mice. We find that repeated stimulation of one pathway at low frequency (0.2 Hz), which does not by itself induce synaptic potentiation, will produce long-lasting synaptic plasticity when paired with a brief conditioning burst applied to an independent afferent pathway. The induction of heterosynaptic L-LTP is associative and critically depends on the precise time interval of pairing: simultaneous, conjunctional pairing induces L-LTP; in contrast, delayed pairing induces short-lasting early-phase LTP. Heterosynaptically induced early-phase LTP could be depotentiated by repeatedly presenting unpaired test stimuli, whereas L-LTP could not. This heterosynaptically induced L-LTP requires PKA and protein synthesis. In addition, heterosynaptically induced L-LTP is impaired in transgenic mice that express KCREB (a dominant negative inhibitor of adenosine 3'5'-cyclic monophosphate response element-binding protein-mediated transcription) in the hippocampus. These mice have previously been shown to be impaired in spatial memory but have normal L-LTP as induced by a conventional homosynaptic tetanic protocol. These data suggest that at least in some instances this L-LTP-inducing protocol may better model behaviorally relevant information storage and the in vivo mechanisms underlying long-lasting memories.

Fig. 1.

Pairing of 0.2-Hz stimuli in a test pathway with conditioning bursts in an independent afferent pathway produces L-LTP at the Schaffer collateral synapse. (A1) Schematic representation of a transverse slice through the hippocampus, showing the positioning of two stimulating electrodes and a single recording electrode in stratum radiatum of the CA1 cell field. (A2) Schematic representation of a single CA1 pyramidal cell, showing the position of two stimulating electrodes and one recording electrode in stratum radiatum. (B) Test pulses alone do not alter synaptic strength. Sixty test pulses were delivered to test pathway S1 at 0.2 Hz, with no stimulation in the conditioning pathway S2, as illustrated (Right). This produced no change in synaptic strength in either the test pathway (Top Right) or the simultaneously recorded conditioning pathway. Test stimuli produced simple postsynaptic EPSPs (Top Right, calibration, 20 ms, 2 mV). Independence of S1 and S2 was defined as the absence of paired-pulse facilitation of S1 by S2 with a 50-ms interval (data not shown). Stimulation in both pathways was at an intensity that produced 30–50% of the maximum postsynaptic response before any potentiation. (C) Asynchronous pairing of 60 test stimuli with conditioning bursts, as illustrated in Inset (S1–S2 interval, 100 ms) produced no potentiation in the test pathway (94 ± 3% at 1 h after pairing, n = 5, Upper) but did produce L-LTP in the simultaneously recorded conditioning pathway (162 ± 12%, n = 5). Conditioning bursts produced a more complex EPSP (B Middle Right); asynchronous pairing produced both waveforms with no apparent summation of the two (B Bottom Right, calibration, 140 ms, 2 mV). (D) After a 30-min stable baseline, conjunctional pairing, with an S1–S2 interval of 0.1 ms, produced L-LTP in both the test pathway (159 ± 13% at 3 h, n = 5, Left) and the conditioning pathway (141 ± 10%, n = 5, Right). This stimulation produced a more complex waveform, in which the S1 and S2 waveforms summated (Left Inset, calibration bars, 2 mV, 30 ms).

Fig. 2.

Pairing-induced L-LTP depends on the NMDA receptor, PKA, and a brief S1–S2 interval. (A) Stimulation and pairing was repeated in the presence of the NMDA receptor blocker APV (50 μM). This did not affect the EPSP to test stimulus S1 but did truncate the late phase of response to conditioning burst S2 (Left Inset, S1 and S2 were presented asynchronously, with a 100-ms S1–S2 interval), indicating that this conditioning burst activates the NMDA receptor. The complex EPSP response to conjunctional pairing of S1 and S2 was both truncated and reduced in amplitude in the presence of APV (Right Inset, calibration, 2 mV, 50 ms). When the conjunctional pairing protocol was applied, as in Fig. 1 A, APV eliminated the production of any early or L-LTP (96 ± 4% at 100 min after pairing, n = 6). (B) VGCCs inhibitor blocked the L-LTP. In the presence of inhibitor nifidipine (15 μM), the LTP at 1 h was not affected, but L-LTP at 3 h was significantly reduced (113 ± 10%, n = 5 for each condition, P < 0.01). (C) Conjunctional pairing again produced L-LTP (upper graph; conditions identical to Fig. 1D; 168 ± 10% baseline 3 h after pairing; n = 6). Delayed pairing, with a S1-S2 interval of 20 ms during pairing but with otherwise identical stimulation, produced only E-LTP (lower graph; 112 ± 16% baseline 3 h after pairing; n = 6). (D) Both of these forms of LTP require PKA. The PKA inhibitor KT5720 (1 μM, perfusion for 2 h) truncated the L-LTP produced by conjunctional pairing to E-LTP [upper, KT5720 (•), 93 ± 5% 3 h after pairing; control (○), 168 ± 11%; n = 7 for each condition; P < 0.01]. KT5720 also reduced the E-LTP produced by delayed pairing [lower, KT5720 (○), 109 ± 14% 40 min after pairing; control, 154 ± 15%; n = 7 for each condition, P < 0.01].

Fig. 3.

Pairing-induced L-LTP depends on protein synthesis and on CREB-mediated transcription. (A) Conjunctional and delayed pairing were performed in the presence of the protein synthesis inhibitor anisomycin (20 μM, applied 20 min before until 90 min after pairing). This truncated L-LTP produced by conjunctional pairing [Upper, anisomycin (•), 118 ± 18% baseline 3 h after pairing; control (○), 168 ± 10%; n = 7 for each condition; P < 0.01]. However, anisomycin had no effect on E-LTP produced by delayed pairing (Lower). (B) Pairing-induced L-LTP, but not high-frequency-induced L-LTP in the conditioning pathway, depends on CREB. Conjunctional pairing was induced in slices from dorsal hippocampus of KCREB-expressing transgenic mice (4) and littermate controls. Early potentiation in the test pathway was not altered in transgenic mice, but L-LTP was reduced [Upper, KCREB (○), 118 ± 6% baseline 3 h after pairing; littermate control (•), 150 ± 11%; n = 8 for each genotype; P < 0.05]. L-LTP was normal in the conditioning pathway (Lower). (Inset) Representative EPSPs before and 3 h after pairing in control (Left) and KCREB (Right) mice. (Calibration, 2 mV, 10 ms.)

Fig. 4.

Pairing-induced E-LTP, but not L-LTP, can be depotentiated by subsequent unpaired test stimuli. (A) Delayed pairing (20 ms) was immediately followed by 60 more unpaired test stimuli to pathway S1 at 0.2 Hz. This treatment truncated and reduced the E-LTP [depotentiated (•), 112 ± 11% 40 min after pairing; control, 147 ± 9%; n = 7 for each condition; P < 0.05]. (Inset) Regular E-LTP (Left) and depotentiated E-LTP (Right) 40 min after pairing (calibration, 2 mV, 10 ms). (B) Depotentiation was ineffective 5 min after delayed pairing; 60 unpaired stimuli were delivered to the test pathway S1 5 min after delayed pairing. E-LTP was not significantly reduced by this treatment (130 ± 6% 40 min after pairing; control is the same as in A; n = 6; P > 0.5). (C) Conjunctional pairing-induced L-LTP cannot be depotentiated by this treatment; 60 unpaired stimuli were applied to test pathway S1 immediately after the induction of L-LTP by conjunctional pairing (0.1 ms). This treatment did not significantly reduce the LTP (n = 4 for each condition).