Src Activation in the Induction of Long-Term Potentiation in CA1 Hippocampal Neurons

See allHide authors and affiliations

Science  27 Feb 1998:
Vol. 279, Issue 5355, pp. 1363-1368
DOI: 10.1126/science.279.5355.1363


Long-term potentiation (LTP) is an activity-dependent strengthening of synaptic efficacy that is considered to be a model of learning and memory. Protein tyrosine phosphorylation is necessary to induce LTP. Here, induction of LTP in CA1 pyramidal cells of rats was prevented by blocking the tyrosine kinase Src, and Src activity was increased by stimulation producing LTP. Directly activating Src in the postsynaptic neuron enhanced excitatory synaptic responses, occluding LTP. Src-induced enhancement of α-amino-3-hydroxy-5-methylisoxazolepropionic acid (AMPA) receptor–mediated synaptic responses required raised intracellular Ca2+ and N-methyl-d-aspartate (NMDA) receptors. Thus, Src activation is necessary and sufficient for inducing LTP and may function by up-regulating NMDA receptors.

Long-term potentiation is a persistent enhancement in the efficacy of synaptic transmission that has been proposed to be a principal cellular substrate underlying learning and memory (1). LTP is induced by a cascade of biochemical steps that, for a main form of LTP, occur in the postsynaptic neuron (2). Protein tyrosine phosphorylation is necessary for induction of LTP (3); however, it has not been determined which of the many tyrosine kinases expressed in the central nervous system (CNS) is essential for LTP induction (4), and the role of the kinase is unclear. Tyrosine phosphorylation regulates the function of NMDA receptors (5), which are necessary for induction of LTP at many synapses (6). The regulation of NMDA receptors by tyrosine phosphorylation is via the nonreceptor protein tyrosine kinase Src (7). Thus, we set out to determine whether Src participates in LTP.

We made whole-cell patch-clamp recordings from pyramidal neurons in the CA1 region of hippocampal slices from rat brains; field potentials were recorded by an extracellular electrode (8). Excitatory synaptic responses were evoked by stimulating the Schaffer collateral inputs to CA1 neurons. LTP at these synapses is known to depend on NMDA receptors (6). In order to determine whether Src is necessary for LTP induction, we made use of a unique domain peptide fragment, Src(40–58), which is known to block Src function (7). Src(40–58) was applied directly into the neurons by diffusional exchange from the patch electrode (Fig.1A). During application of Src(40–58), tetanic stimulation caused short-term but not long-lasting potentiation of the intracellularly recorded excitatory postsynaptic potentials (EPSPs): the slope of the EPSPs was 99 ± 5.7% (mean ± SEM) of the baseline level by 30 min after tetanic stimulation (n = 6 cells). However, the tetanic stimulation did produce a long-lasting increase in field EPSP slope to 182 ± 24% of baseline. Thus, Src(40–58) prevented induction of LTP in the cells in which it was administered intracellularly, but not in neighboring cells. A peptide with the same amino acid composition, but in random order, scrambled Src(40–58), served as a control (9), and did not prevent induction of LTP. Intracellular application of the antibody anti-Src1, which specifically blocks Src action (10), caused the EPSP slope to decline to 120 ± 9.5% of baseline by 30 min after tetanus (n = 7 cells; Fig.1B), whereas the field EPSP slope was at a sustained level of 189 ± 21% of baseline. In contrast, a nonspecific immunoglobulin G (IgG) fraction did not affect LTP induction. With administration of Src(40–58) or of anti-Src1, the tetanic stimulation produced posttetanic potentiation, the peak of which was not different from that of the respective controls. Thus, Src was necessary for induction of LTP.

Figure 1

Blocking Src prevents induction of LTP. (A) Src(40–58) prevented tetanus-induced LTP. (Top) A plot of EPSP slope from representative cells with intracellular administration of Src(40–58) (○) or scrambled Src(40–58) [sSrc(40–58), •]. (Middle) Averaged EPSP slope for experiments with Src(40–58) (○, n = 6) or sSrc(40–58) (•, n = 5). Data were normalized to baseline values (8). (Bottom) Averaged normalized field EPSP (fEPSP) slope with Src(40–58) (○) or sSrc(40–58) (•). (B) Anti-Src1 inhibited tetanus-induced LTP. (Top) EPSP slope plotted for example cells in which anti-Src1 (○) or a nonspecific IgG fraction (•) was administered. (Middle) Averaged normalized EPSP slope for recordings with application of anti-Src1 (○, n = 7) or a nonspecific IgG fraction (•, n = 4). (Bottom) Averaged normalized fEPSP slope with anti-Src1 (○) or nonspecific IgG (•). (C) Src(40–58) had no effect on basal NMDAR EPSCs. (Top) Traces from a cell with intracellular administration of Src(40–58) and bath application of CNQX. Each trace is the average of three EPSCs evoked at membrane potentials from –80 to +60 mV, in 20-mV increments. The I-V relation was determined during the first 5-min period (left) or 30 min after the start of recording (right). (Bottom) I-V relation for averaged amplitudes of NMDAR EPSCs (n = 3) during the first 5 min (○) and the period from 30 to 35 min (•). (D) Src(40–58) had no effect on basal AMPAR EPSCs. (Top) EPSCs from a cell recorded with administration of Src(40–58) but without CNQX. Each trace is the average of two responses evoked at membrane potentials from –80 to +60 mV, from the first 5 min (left) or at 30 min (right) after the start of recording. (Bottom) AveragedI-V relation for peak amplitude of AMPAR EPSCs (n = 3) from the first 5 min (○) or for 30 to 35 min (•). In Figs. through 4, horizontal bars above the graphs indicate the periods of peptide administration and the arrow indicates the time of tetanic stimulation. Error bars are ±SEM, and the traces to the right of the graphs are averages of three sweeps taken at the time points indicated by the letters below the x axis.

In the adult CNS, there is a basal level of Src activity (4), and in CA1 neurons this might produce a tonic enhancement of NMDA channel function. Then the blockade of LTP might have been through abolishing ongoing enhancement of NMDA channels, in which case blocking Src would be expected to reduce basal synaptic NMDA responses. However, in voltage-clamp experiments (11), administration of Src(40–58) had no effect on pharmacologically isolated NMDA receptor–mediated excitatory postsynaptic currents (NMDAR EPSCs) (Fig. 1C). Moreover, Src(40–58) had no effect on AMPA receptor–mediated (AMPAR) EPSCs (Fig. 1D). Thus, neither synaptic NMDARs nor synaptic AMPARs were enhanced tonically by basal Src function.

We hypothesized that Src might be activated during LTP induction. To test this, we measured Src catalytic activity by means of an immune-complex kinase assay (12). The 32P incorporation produced by Src immunopurified from slices that had received tetanic stimulation was greater than that from control slices that had received only test stimulation (Fig.2, A and B). Because there was no difference in the level of Src protein (Fig. 2A), tetanus caused an increase in Src activity.

Figure 2

Src activity is increased by tetanic stimulation, and activation of endogenous Src causes synaptic potentiation. (A) Immune complex kinase assays were done with Src immunoprecipitated after control stimulation or 1 min or 5 min after tetanus. Proteins were separated by SDS-PAGE and transferred to nitrocellulose, and 32P was detected by exposure to a Phosphor Screen. Results of a representative immune complex kinase experiment are shown; phosphorylation of Src or of enolase is indicated by the arrows. The corresponding anti-Src immunoblot is shown below. (B) 32P labeling was quantified for Src (solid bars) and enolase (open bars). Data from 1- and 5-min times were normalized as a percentage of control (n = 4 experiments). (C) (Top) A plot of EPSP slope during intracellular application of pp60c-Src (•) or heat-inactivated (boiled) pp60c-Src (○). (Bottom) Averaged EPSP slope with application of pp60c-Src(n = 7) or boiled pp60c-Src(n = 5). (D) (Top) A plot of EPSP slope during intracellular application of EPQ(pY)EEIPIA (1 mM, •), or EPQYEEIPIA (1 mM, ○). Neither EPQ(pY)EEIPIA nor EPQYEEIPIA affected resting membrane potential or input resistance (not shown in figure). (Bottom) Averaged EPSP slope during application of EPQ(pY)EEIPIA (n = 8) or EPQYEEIPIA (n= 4). (E) (Top) A record of EPSP slope during intracellular application of Src(40–58) (○) or sSrc(40–58) (•). During the period indicated by the upper horizontal line, EPQ(pY)EEIPIA was administered by perfusion of the patch electrode. (Bottom) Averaged effects of EPQ(pY)EEIPIA on EPSP slope during application of Src(40–58) (○, n = 6) or sSrc(40–58) (•, n = 5). (F)I-V relations for EPSCs evoked during application of EPQ(pY)EEIPIA. EPSCs were evoked at membrane potentials from –80 to +60 mV. (Top) The records are superimposed EPSC traces collected during the first 5 or 30 min after the start of recording. (Bottom) Averaged peak amplitudes of AMPAR EPSCs are plotted in the graph (n = 4 cells) for the first 5 min (•) or at 30 min (○).

To determine whether increasing Src activity affects synaptic responses, we administered exogenous recombinant Src (pp60c-Src), which was found to increase EPSP slope to 185 ± 24% of baseline (n = 7 cells). In contrast, heat-inactivated pp60c-Src had no effect (Fig. 2C). To examine the effect of activating endogenous Src, we used the high-affinity peptide EPQ(pY)EEIPIA (13), which is an activator of tyrosine kinases in the Src family (14). EPQ(pY)EEIPIA was applied alone or with Src(40–58) to determine whether the effects required Src itself. Application of EPQ(pY)EEIPIA produced an increase in EPSP slope to a sustained level at 226 ± 22% of baseline (n = 8 cells; Fig. 2D). On the other hand, the nonphosphorylated form of the peptide, EPQYEEIPIA, which does not activate tyrosine kinases (14), did not affect EPSPs (n = 4 cells). Moreover, during administration of Src(40–58), perfusion with EPQ(pY)EEIPIA (15) had no effect on EPSP slope (n = 6 cells, Fig. 2E). But during administration of scrambled Src(40–58), perfusion of EPQ(pY)EEIPIA did produce an increase in EPSP slope to 200 ± 19% of baseline (n = 5 cells). Thus, endogenous Src was necessary for the enhancement of EPSPs by EPQ(pY)EEIPIA. In voltage-clamp recordings, EPQ(pY)EEIPIA potentiated AMPAR EPSCs through an increase in AMPAR EPSC conductance, with no change in driving force (Fig. 2F). Overall, activation of Src was sufficient to enhance EPSPs.

If endogenous Src participates in LTP produced by tetanic stimulation, then LTP and the enhancement by the activating peptide may occlude each other. This was investigated by applying EPQ(pY)EEIPIA, and when the EPSPs had been maximally enhanced tetanic stimulation was delivered (Fig. 3A). This stimulation caused posttetanic potentiation but produced no long-lasting increase in EPSP slope. In contrast, when EPQYEEIPIA was administered, delivering tetanic stimulation at the same time after beginning the recording caused a long-lasting increase in EPSP slope (212 ± 5.3% of baseline, n = 5 cells). In other experiments, tetanus produced a lasting potentiation of EPSP slope (212 ± 28% of baseline, n = 5 cells), but there was no further increase when EPQ(pY)EEIPIA was applied intracellularly (Fig. 3B). On the other hand, in cells not conditioned by tetanic stimulation, perfusion of EPQ(pY)EEIPIA at the same time after beginning recording caused a progressive enhancement of EPSP slope that reached a stable level at 225 ± 23% of baseline (n = 4 cells). Thus, Src-induced enhancement of EPSPs and tetanus-induced LTP were mutually occluded.

Figure 3

Src-induced potentiation and LTP occlude each other. (A) Effect of tetanic stimulation (arrow) during intracellular application of EPQ(pY)EEIPIA (1 mM, •) or EPQYEEIPIA (1 mM, ○). Recordings of EPSP slope during application of EPQ(pY)- EEIPIA (n = 5) or EPQYEEIPIA (n = 5) from representative cells are plotted at the top. The averaged EPSP slope is shown in the graph in the bottom. (B) Effect of tetanus (arrow) on action of EPQ(pY) EEIPIA. Representative recordings of EPSP slope when tetanus was delivered 10 min after the start of recording (•) or without tetanus (○) are shown in the top graph. EPQ(pY)EEIPIA (1 mM) was actively perfused during the period indicated by the horizontal bar above the top graph. The averaged EPSP slope is plotted in the bottom graph (n = 5 with tetanus, n = 4 without tetanus). Neither EPQ(pY)EEIPIA nor EPQYEEIPIA affected the tetanus-induced potentiation of the fEPSP slope.

Because LTP in CA1 neurons depends on raising the intracellular concentration of Ca2+ (16), we next determined whether the Src-induced enhancement of AMPAR EPSCs might be linked to a rise in [Ca2+]. In previous experiments, AMPA receptors appeared not to be regulated by Src (7), but in those experiments, in contrast to the present ones, a high level of intracellular Ca2+ buffering was used. To determine whether Src-induced enhancement of AMPAR EPSCs requires raised intracellular [Ca2+], we increased the buffering capacity of the intracellular solution (17). With the high Ca2+-buffering solution, EPQ(pY)EEIPIA had no effect on AMPAR EPSCs (Fig. 4A). EPQ(pY)EEIPIA remains active in the high Ca2+-buffering solution because there was an increase in NMDAR EPSCs (Fig. 4, A and B). The increase in NMDAR EPSCs was associated with no change in driving force or in current-voltage (I-V) relationship (Fig. 4B) and was produced with low Ca2+-buffering solution (18). Also, the enhancement of NMDA currents by EPQ(pY)EEIPIA was prevented by Src(40–58) and was not produced by EPQYEEIPIA. Thus, the Src-induced increase in AMPAR EPSCs was Ca2+-dependent, but the potentiation of NMDAR EPSCs was Ca2+-independent.

Figure 4

Src-induced enhancement of synaptic AMPAR responses depends on raising intracellular [Ca2+] and on NMDARs. (A) Potentiation of AMPAR EPSCs is Ca2+-dependent, but potentiation of the NMDAR EPSC component is Ca2+-independent. (Top) Plot of amplitude of AMPAR EPSCs (•) and the NMDAR component (○) during a recording with high Ca2+-buffering intracellular solution and application of EPQ(pY)EEIPIA. The short bar indicates the period of bath application of MK-801 (10 μM). The NMDAR component was measured 100 ms after the stimulation. (Bottom) Effect of EPQ(pY)- EEIPIA on averaged AMPAR ESPCs or the NMDAR component during recordings with high Ca2+-buffering intracellular solution (n = 5). Data were normalized to EPSCs in the first minute of recording. (B) I-Vrelation for pharmacologically isolated NMDAR EPSCs during application of EPQ(pY)EEIPIA. (Top) Superimposed NMDAR EPSCs evoked at membrane potentials from –80 to +60 mV. Traces are from the first 5 min and after 30 min of recording. (Bottom) Averaged amplitudes of NMDAR EPSCs (n = 3 cells) from the first 5 min (•) or after 30 min (○). (C) MK-801 (10 μM) was bath-applied just before the start of recording with intracellular solution containing EPQ(pY)EEIPIA. The upper graph shows a representative plot of EPSP slope from one cell, and the lower graph is averaged normalized data (n = 4). (D) EPQ(pY)EEIPIA was administered intracellularly and MK-801 (10 μM) was bath-applied during the period indicated by the short horizontal line, after the potentiation of EPSP slope was established. One example is shown in the upper graph and the averaged normalized EPSP slope is plotted below (n = 5).

Because Src is associated with and up-regulates the function of NMDA receptors (7), we questioned whether NMDA receptors are required for Src-induced enhancement of AMPAR EPSCs. We blocked NMDA receptors by bath-applying the antagonist MK-801 during experiments with low Ca2+-buffering intracellular solution. When MK-801 was applied starting just before whole-cell recording, administration of EPQ(pY)EEIPIA produced no change in EPSP slope (Fig. 4C). In other experiments, after synaptic responses had been potentiated by EPQ(pY)EEIPIA, MK-801 had no effect on AMPAR EPSPs (Fig. 4D). Thus, NMDA receptor activation was necessary to induce, but not to sustain, the Src-induced potentiation of AMPAR-mediated synaptic responses.

Here, blockade of Src prevented induction of LTP and activation of Src, or administration of recombinant Src, induced lasting potentiation that occluded LTP induction. Like tetanus-induced LTP, the potentiation produced by directly activating Src depended on a rise in intracellular [Ca2+] and on NMDARs. Thus, Src fulfills necessary and sufficient conditions to be considered a mediator of LTP induction at Schaffer collateral CA1 synapses. Although there is a basal level of Src function, this did not appear to contribute to LTP, but rather the activation of Src as a consequence of tetanic stimulation was required to induce LTP (19). Thus, activation of Src provides a biochemical mechanism for gating induction of LTP.

Models of biochemical events underlying induction of LTP in hippocampal CA1 neurons focus on the signaling cascades initiated by Ca2+ influx through NMDARs (16, 20). The most parsimonious explanation for our findings is that during induction of LTP, Src is rapidly activated, which leads to enhanced NMDAR function. Enhancing NMDAR function results in increased Ca2+ entry, which may trigger the downstream signaling cascade. Hence, the present results indicate that for LTP induction there is a hitherto unexpected step upstream of NMDARs (21). Src is widely expressed in the nervous system (22), and thus Src may have a common role in the plasticity of excitatory synaptic transmission in many regions of the CNS.

  • * To whom correspondence should be addressed. E-mail: roder{at}


View Abstract

Navigate This Article