N-Methyl-d-aspartate receptor (NMDAR)–mediated synaptic responses in hippocampal CA1 pyramidal cells are depressed during NMDAR-dependent long-term depression (LTD) due to mechanisms, in part, distinct from those underlying LTD of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR)–mediated synaptic responses. The mechanisms underlying dedepression of synaptic NMDARs, however, are not known. We find that dedepression of NMDAR-mediated synaptic responses in the CA1 region of the rat hippocampus is input specific and does not require synaptic stimulation to be maintained. The induction of dedepression does not require activation of metabotropic glutamate receptors, L-type Ca2+ channels, or release of Ca2+ from intracellular stores. It does, however, rely on activation of NMDARs. In contrast to the dedepression of AMPAR-mediated synaptic responses, dedepression of NMDAR-mediated synaptic responses does not depend on activation of calcium/calmodulin-dependent protein kinase II, protein kinase C, cAMP-dependent protein kinase, or Src kinases. However, dedepression of synaptic NMDARs is significantly impaired by inhibitors of mitogen-activated protein kinase signaling. Specifically, inhibitors of extracellular signal-regulated kinase 1/2 prevented normal dedepression of synaptic NMDARs by a mechanism that did not require protein synthesis. These results provide further evidence that synaptic NMDARs can be bidirectionally modified by activity but by mechanisms distinct from those responsible for the activity-dependent, bidirectional modulation of synaptic AMPARs.
Activity-dependent, bidirectional control of synaptic strength, exemplified by various forms of long-term potentiation (LTP) and long-term depression (LTD), has been extensively studied at mammalian CNS synapses and is thought to contribute to many forms of experience-dependent plasticity, including learning and memory (Malenka and Bear 2004). Although most excitatory synapses contain both α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) and N-methyl-d-aspartate receptors (NMDARs), because AMPARs are thought to be the predominant source of basal synaptic responses, almost all work on the mechanisms underlying LTP and LTD has focused on the modulation of AMPAR-mediated synaptic responses. Clearly, however, the bidirectional modulation of NMDAR-mediated synaptic responses has important functional implications for NMDAR-dependent forms of synaptic plasticity. For example, limiting the number of activatable synaptic NMDARs by pharmacological methods can either block LTP (Collingridge et al. 1983; Morris et al. 1986; Wigstrom and Gustafsson 1984) or elicit LTD of AMPAR-mediated responses by an induction protocol that would otherwise elicit LTP (Cummings et al. 1996). Moreover, alterations in NMDAR function that have no detectable effect on AMPAR-mediated synaptic transmission may play an important role in metaplasticity at Hebbian synapses (Abraham and Bear 1996; Huang et al. 1992). Therefore the functional state and number of synaptic NMDARs constitute one important factor influencing the triggering of NMDAR-dependent forms of LTP and LTD.
In the context of the functional role of NMDARs, a critical question is whether, like synaptic AMPARs, they are bidirectionally modifiable by synaptic activity. There has been considerable debate whether the induction protocols that elicit LTP of AMPAR-mediated synaptic responses also cause LTP of NMDAR-mediated synaptic responses with some laboratories observing such plasticity (Aniksztejn and Ben-Ari 1995; Bashir et al. 1991; Grosshans et al. 2002; Kullmann et al. 1996; Watt et al. 2004; Xiao et al. 1995) and others not (Durand et al. 1996; Kauer et al. 1988; Liao et al. 1995; Montgomery et al. 2001; Muller and Lynch 1988). In contrast, LTD of NMDAR-mediated synaptic responses has been observed consistently in response to induction protocols that elicit NMDAR-dependent LTD of AMPAR responses (Gean and Lin 1993; Montgomery and Madison 2002; Morishita et al. 2005; Selig et al. 1995; Sobczyk and Svoboda 2007; Xiao et al. 1994; Xie et al. 1992). Moreover, LTD of NMDAR-mediated synaptic responses can be induced with stimuli that do not cause LTD of AMPAR-mediated synaptic responses (Selig et al. 1995) and the signal transduction mechanisms underlying these two forms of LTD appear to be different (Morishita et al. 2005; but see Montgomery et al. 2005).
The ability to consistently trigger LTD of both NMDAR- and AMPAR-mediated synaptic responses and the fact that the mechanisms underlying these forms of plasticity appear to differ raises the question of whether the mechanisms underlying the potentiation or “dedepression” of these depressed synaptic responses also differ. Only a small number of studies have examined dedepression of AMPAR-mediated synaptic responses and the mechanisms underlying this potentiation of depressed responses appear to differ from those responsible for standard LTP of basal AMPAR-mediated synaptic transmission (Daw et al. 2000; Lee et al. 2000). Dedepression of NMDAR-mediated responses has been observed (Selig et al. 1995; Xiao et al. 1995), although nothing is known about mechanisms underlying this form of plasticity.
In the present study we examined whether dedepression can be induced after LTD of the NMDAR-mediated field excitatory postsynaptic potential (NMDAR fEPSP) in the CA1 region of the rat hippocampus. We find that dedepression of the NMDAR fEPSP is input specific and triggered by strong activation of NMDARs but does not require synaptic activity to be maintained. Moreover, this form of plasticity was not blocked by inhibitors of Ca2+/calmodulin-dependent kinase II (CaMKII), protein kinase C (PKC), protein kinase A (PKA), or nonreceptor tyrosine (Src) kinases. The dedepression of the NMDAR fEPSP, however, was impaired by inhibiting mitogen-activated protein kinases (MAPKs) by a mechanism that did not require protein synthesis. Directly loading CA1 pyramidal cells with a MAPK inhibitor also impaired the dedepression of NMDAR-mediated excitatory postsynaptic currents (NMDAR EPSCs), suggesting that postsynaptic MAPK activity is required for this form of plasticity. These results demonstrate that NMDAR-mediated synaptic responses are bidirectionally modifiable by activity but that the mechanisms underlying this plasticity are likely different from those controlling the bidirectional control of AMPAR-mediated synaptic responses.
Hippocampal slices (300 μm thick) were prepared from 3- to 4-wk-old Sprague–Dawley rats as previously described (Morishita et al. 2001). Briefly, transverse slices from the dorsal portion of the hippocampus were cut in ice-cold sucrose solution containing (in mM): sucrose 238, KCl 2.5, NaH2PO4 1.3, NaHCO3 26.2, CaCl2 1, MgSO4 2, and d-glucose 11 (saturated with 95% O2-5% CO2). Slices were transferred to a holding chamber containing external solution consisting of (in mM): NaCl 119, KCl 2.5, NaH2PO4 1, NaHCO3 26.2, CaCl2 2.5, MgSO4 1.3, and d-glucose 11 (saturated with 95% O2-5% CO2). Slices were allowed to equilibrate at room temperature for ≥1.5 h before being transferred to a recording chamber and perfused (2 ml/min) with oxygenated external solution warmed to 28–30°C. For experiments in which slices were soaked with different inhibitors, after a 1-h recovery period, slices were transferred to a holding chamber containing inhibitor and incubated for an additional 1.5 h before placement into a recording chamber that was perfused with external solution containing the same concentration of inhibitor.
Field and whole cell recordings were made in area CA1. NMDAR-mediated dendritic field EPSPs (NMDAR fEPSPs) were pharmacologically isolated by lowering the external Mg2+ concentration to 0.5 mM and adding 2,3-dihydroxy-6-nitro-7-sulfamoyl-benzo[f]quinoxaline-2,3-dione (NBQX, 5 μM) and picrotoxin (50 μM) to block synaptic transmission mediated by AMPARs and γ-aminobutyric acid type A receptors, respectively. NMDAR fEPSPs were recorded with a patch pipette filled with 1 M NaCl solution containing 10 mM HEPES (adjusted to pH 7.4 with NaOH). Basal NMDAR fEPSPs were evoked at 0.05 Hz with a monopolar stimulating electrode positioned in the stratum radiatum. Two pathway experiments were conducted by placing a second electrode in the stratum radiatum on the opposite side of the recording electrode and slightly staggered relative to the position of the first stimulating electrode. Slices were stimulated for ≥10 min before data acquisition. During this “warm-up” period stimulation intensity was adjusted to evoke stable baseline NMDAR fEPSPs. LTD was elicited by two 5-Hz, 3-min trains each separated by 4 min. Dedepression was induced with two 100-Hz, 1-s trains separated by 20 s. Whole cell recordings were made with patch pipettes filled with a solution containing (in mM): CsMeSO4 117.5, CsCl 15.5, HEPES 10, NaCl 8, TEA-Cl 10, MgCl2 1, Mg-ATP 4, Na-GTP 0.3, U0124 0.020, or U0126 0.020 (300–310 mOsM, pH adjusted to 7.3 with CsOH, 4–7 MΩ). Cells were voltage clamped at −65 mV. For experiments involving simultaneous whole cell and field recordings (⇓⇓⇓⇓⇓Fig. 6) the patch pipette was positioned “in-line” above the field recording electrode to maximize the likelihood of recording from a neuron that contributed to the population fEPSP response. The fEPSP was continuously monitored while attempting to make a whole cell recording. During this period the positive pressure administered to the patch pipette and resulting ejection of the Cs-based internal solution from the patch pipette into the pyramidal layer caused the fEPSP to diminish in size. As a result whole cell access was not made until the fEPSP recovered to “prepatch” levels. Time for the fEPSP to recover varied from experiment to experiment, depending on whether multiple attempts were made to attain a whole cell recording from a pyramidal neuron. After whole cell access ≥10 min elapsed before applying the dedepression induction protocol to allow for diffusion of drug into the cell. Total diffusion times did not differ by more than 3 min for each drug. During the tetanic stimulation used to elicit dedepression, whole cell recordings were switched from voltage clamp to current clamp. Input- and series resistances were calculated from the current arising from a −4 mV rectangular pulse lasting 70 ms. Cells were not included in the data analysis if series resistance varied by >20% throughout the experiment. Field and whole cell recordings were filtered at 1 and 2 kHz, respectively, and digitized at 5 kHz with an A/D board (National Instruments) driven by custom acquisition software designed to run on IGOR Pro (version 5.04).
Drugs and inhibitors
Drugs that could not be dissolved in dH2O were dissolved in DMSO. Stock concentrations of drugs were prepared at 500–1,000 × the final concentrations used in the study. NBQX, d-2-amino-5-phosphonovaleric acid (d-APV), nifedipine, thapsigargin, (2S,1′S,2′S)-2-(2-caboxycyclopropyl)-2-(9H-xanthen-9-yl)glycine (LY341495), bisindolylmaleimide I (Bis), 3-(4-chlorophenyl)1-(1,1-dimethylethyl)-1H-pyrazolo[3,4-d]pyrimidin-4-amine (PP2), KN-93, U0124, U0126, and 2′-amino-3′-methoxyflavone (PD98059) were purchased from Tocris. Picrotoxin, H-89, and cycloheximide were purchased from Sigma. Additional supplies of KN-93 and PD98059 were purchased from Calbiochem.
Dedepression graphs were generated by averaging the initial slope of the NMDAR fEPSPs or peak amplitude of the NMDAR EPSCs in 1-min bins and expressing the data as a percentage of the averaged 10-min baseline collected before LTD induction. Initial slopes of NMDAR fEPSPs were calculated using a 2-ms time window beginning at a fixed time point that was one third of the peak amplitude of the baseline responses. Once the cursors defining the 2-ms window were set, they were not changed during the course of the experiment. NMDAR EPSC peak amplitudes were calculated by averaging all points within a 3-ms time window on the baseline of each trace and subtracting this from the peak amplitude that was also measured using a 3-ms time window. Unless otherwise stated, dedepression percentage values in the text and in Fig. 7 represent averaged data collected between 35 and 40 min after the dedepression induction protocol. These values were compared with averaged data collected between 5 and 10 min of the original baseline before LTD induction. Data are presented as means ± SE. To determine whether synaptic responses after dedepression were significantly different from baseline responses for single experimental manipulations, a paired t-test was used (P < 0.01). A Mann–Whitney U test was used (P < 0.05) when data were compared between two independent experimental manipulations (Fig. 6).
Dedepression of NMDAR-mediated synaptic transmission
To determine whether dedepression of NMDAR-mediated synaptic responses occurs at Schaffer collateral/commissural synapses in the CA1 region of the hippocampus, LTD of NMDAR fEPSPs was first induced with two 5-Hz, 3-min trains (Fig. 1). This stimulation induced LTD (25–50% depression) of NMDAR fEPSPs in all slices tested (n = 127). Previous studies demonstrated that a dedepression of AMPAR-mediated synaptic responses could be induced after high-frequency stimulation (HFS) of the afferent inputs (Daw et al. 2000; Lee et al. 2000; Mulkey and Malenka 1992; Selig et al. 1995). We therefore tested whether a similar HFS could also induce a dedepression of synaptic NMDARs and found that HFS (two 100-Hz, 1-s trains) induced a robust dedepression of the NMDAR fEPSP (post-HFS: 103 ± 3.1%, n = 20; Figs. 1A2 and 7). The time course of the dedepression was often not immediate but gradual and commonly took 15–30 min to fully develop (Fig. 1, A2, C, and D). The absolute magnitude of the dedepression (normalized to a 5-min baseline collected just before delivery of the HFS) was 159 ± 4.5% (Fig. 1A2, n = 20). In contrast, if HFS was given alone to basal NMDAR fEPSPs the magnitude of the potentiation was 111 ± 7.9%, which was not significantly different from baseline (Fig. 1B, n = 5, P = 0.09). The dedepression of NMDAR fEPSPs was also input specific because HFS applied to one of two pathways in which LTD had been elicited induced dedepression only in that pathway and not the other (Fig. 1C, n = 8). These experiments also indicate that without HFS the LTD of NMDAR fEPSPS is stable, lasting for the full duration of the recordings (∼70–80 min after LTD induction; Fig. 1C, control path). The changes in NMDAR fEPSPs during LTD and dedepression were not accompanied by changes in fiber volley shape or amplitude (see sample traces in Figs. 1 and 4), indicating that activity-dependent changes in axonal excitability do not contribute to these forms of synaptic plasticity.
To determine whether the dedepression required maintained synaptic activity HFSs were delivered to both pathways after LTD and stimulation was then stopped temporarily in one pathway while the development of dedepression in response to basal stimulation was monitored in the other pathway (Fig. 1D, n = 5). Resumption of stimulation of the test pathway revealed that dedepression still occurred despite the cessation of stimulation for 30 min after the HFS. Thus dedepression of NMDAR fEPSPs does not require activity after its induction. This result was not simply a consequence of stopping afferent stimulation because LTD of the NMDAR fEPSP, not dedepression, was observed if basal stimulation was stopped for the same duration but in the absence of HFS (Fig. 1E, n = 6). The transient enhancement of the NMDAR fEPSP that occurred after basal stimulation was restarted in both sets of experiments (Fig. 1, D and E) is due to the cessation of afferent stimulation for 30 min because a similar transient enhancement of basal AMPAR fEPSPs was observed after stopping stimulation for 30 min (n = 3; data not shown).
Induction of dedepression requires NMDAR activation
To examine the mechanisms responsible for triggering dedepression of NMDAR fEPSPs we applied well-established pharmacological inhibitors of receptors or signaling proteins while monitoring NMDAR fEPSPs in two independent pathways. The control pathway served to ensure that 1) the drugs had no significant effects on basal synaptic transmission and 2) the preparations were stable. We first examined whether induction of dedepression of NMDAR fEPSPs required activation of NMDARs by applying d-APV (50 μM) during the HFS (Fig. 2A). This blocked the dedepression normally induced by HFS, as evidenced by the fact that LTD was still present after washout of the drug (pre d-APV: 67 ± 1.2%; post d-APV: 71 ± 2.2%, n = 8). The return of NMDAR fEPSPs to their original baseline value in the control paths in these experiments indicates that the continued presence of LTD (and lack of dedepression) was not attributable to the incomplete washout of the d-APV (Fig. 2A). Furthermore, repeating the HFS after d-APV washout induced dedepression (post-HFS: 110 ± 2.8%; Fig. 2A).
Because activation of metabotropic glutamate receptors (mGluRs) alone (Bashir et al. 1993; Tyszkiewicz et al. 2004) or in conjunction with NMDARs (Kotecha et al. 2003) has been reported to potentiate NMDAR-mediated synaptic responses we next examined the requirement for mGluRs in dedepression of NMDAR fEPSPs by applying the general (at the concentration used) mGluR antagonist, LY341495 (100 μM) during the HFS. This manipulation did not prevent dedepression after HFS (113 ± 2.1%, n = 8, Fig. 2B), indicating that the dedepression of NMDAR fEPSPs does not require activation of mGluRs. To test the role of other potential sources of Ca2+ that may be required for the HFS to trigger dedepression, we applied the L-type Ca2+ channel antagonist nifedipine (25 μM) and thapsigargin (5 μM) to block release of Ca2+ from intracellular stores. Neither of these manipulations prevented the dedepression of NMDAR fEPSPs (Fig. 2, C and D) (nifedipine, 110 ± 3.7%, n = 5; thapsigargin, 107 ± 1.5%, n = 7). Together these results suggest that strong activation of NMDARs alone is sufficient to trigger dedepression of NMDAR-mediated synaptic responses.
Dedepression does not require CaMKII, PKC, PKA, or Src activity
A number of different protein kinases have been suggested to be involved not only in LTP (see review by Soderling and Derkach 2000) but also in dedepression of AMPAR-mediated synaptic transmission (Daw et al. 2000; Lee et al. 2000). Most prominent among these are CaMKII, PKC, PKA, and Src. Moreover, several of these have been suggested to form complexes with NMDARs and regulate NMDAR function (Wang and Salter 1994; Yu and Salter 1999). To test whether these kinases play a role in the dedepression of NMDAR fEPSPs we presoaked and then perfused slices with a well-established inhibitor of each of these proteins. Each drug was chosen because it was previously reported in several publications to block a form of synaptic plasticity in hippocampal slices at the concentration used in our experiments. Surprisingly, dedepression was still induced in the presence of the CaMKII inhibitor KN-93 (10 μM) (98 ± 4.8%, n = 6, Fig. 3A), the PKC inhibitor BIS (2 μM) (94 ± 7.1%, n = 6, Fig. 3B), the PKA inhibitor H-89 (10 μM) (98 ± 5.8%, n = 8, Fig. 3C), or the Src inhibitor PP2 (5 μM) (98 ± 5.5%, n = 6, Fig. 3D). Thus dedepression of NMDAR-mediated synaptic responses does not appear to require CaMKII, PKC, PKA, or Src activity.
Dedepression involves activation of MAPKs
Another family of protein kinases implicated in LTP of AMPAR-mediated synaptic transmission are activated by the MAPK cascade, specifically the extracellular signal-regulated kinases 1 and 2 (ERK1 and ERK2; also known as p44 and p42 MAPK, respectively) (for reviews see Sweatt 2004; Thomas and Huganir 2004; Wang et al. 2007). Activation of NMDARs by LTP-inducing HFS induces phosphorylation of MAPKs in hippocampal CA1 neurons (Atkins et al. 1998; Dudek and Fields 2001; English and Sweatt 1996, 1997; Opazo et al. 2003), whereas pharmacological inhibition of MAPKs has been reported to block the induction of hippocampal LTP (Atkins et al. 1998; Davis et al. 2000; English and Sweatt 1996, 1997; Opazo et al. 2003; Watabe et al. 2000; Wu et al. 2006). To test the role of MAPK signaling in dedepression of NMDAR-mediated synaptic responses we presoaked and perfused slices with the ERK1/2 inhibitor PD98059 (40 μM). This manipulation significantly reduced the magnitude of the dedepression of the NMDAR fEPSP (75 ± 7.9%, n = 9, Fig. 4A) . A structurally unrelated inhibitor of ERK1/2, U0126 (10 μM), also significantly reduced dedepression (76 ± 2.1%, n = 9, Fig. 4B). In contrast, normal dedepression of NMDAR fEPSPs was generated in slices exposed to U0124 (10 μM), an inactive analogue of U0126 (92 ± 1.4% n = 8, Fig. 4C).
Several studies have reported that ERK activation can strongly regulate translation and subsequent protein synthesis (Banko et al. 2005; Giovannini et al. 2001; Kelleher et al. 2004). To determine whether dedepression of NMDAR-mediated synaptic responses involved protein synthesis we incubated and perfused slices with the protein synthesis inhibitor cycloheximide (60 μM). This manipulation, however, failed to block dedepression of the NMDAR fEPSP (112 ± 6.7%, n = 4, Fig. 4D). These results suggest that although ERK1/2 activity is required for dedepression of NMDAR fEPSPs it does so by a mechanism that does not require protein synthesis.
Dedepression involves activation of postsynaptic MAPKs
Because the MAPK inhibitors were applied extracellularly, it is impossible to determine whether the impairment of dedepression was due to pre- or postsynaptic effects of MAPK inhibition. An obvious way to test whether MAPK inhibitors exerted their effects postsynaptically would be to make a whole cell recording from a CA1 pyramidal cell with a pipette solution containing a MAPK inhibitor and administer the dedepression stimulus after LTD of the NMDAR-mediated EPSC. Unfortunately, we could not use this approach because, after eliciting LTD of the NMDAR-mediated EPSC, dedepression could not be induced (n = 3; data not shown), presumably due to “washout” of the sort that makes obtaining LTP of AMPAR-mediated EPSCs very difficult after >10 min of whole cell recording (e.g., Malinow and Tsien 1990). Therefore we developed a protocol that allowed us to obtain whole cell recordings from cells that had previously undergone LTD. Specifically, we induced LTD of NMDAR fEPSPs by bath application of NMDA (10 μM) (77 ± 4.7%, n = 6, Fig. 5A), a protocol previously used to evoke LTD of AMPAR-mediated synaptic responses (Kameyama et al. 1998; Lee et al. 1998). This chemically induced LTD (chem LTD) occluded synaptically evoked LTD of NMDAR fEPSPs (Fig. 5A) and, conversely, was occluded by the prior generation of synaptically evoked LTD (n = 7, Fig. 5B). Results of these occlusion experiments suggest that the LTD of NMDAR-mediated synaptic responses generated by these two different induction protocols share underlying mechanisms.
Because the chem LTD protocol presumably induces LTD at all or the majority of synapses on all the CA1 pyramidal cells in the slice preparation, after the chem LTD of NMDAR fEPSPs stabilized, we could then make a whole cell recording and generate dedepression of NMDAR EPSCs (Fig. 6) . During these experiments, the whole cell pipette solutions contained either U0126 (20 μM) to inhibit postsynaptic MAPK activity (n = 6, Fig. 6A) or its inactive analogue, U0124 (n = 6, 20 μM) (Fig. 6B). Cells that were loaded with U0126 exhibited significantly less dedepression of the NMDAR EPSC than that in cells that were loaded with U0124 (P < 0.05, Fig. 6C), a finding suggesting that postsynaptic MAPK activity is required for dedepression of NMDAR-mediated synaptic responses.
We have demonstrated that after LTD of NMDAR-mediated synaptic responses HFS can potentiate these responses back to the pre-LTD baseline. This phenomenon, termed dedepression, is input specific and does not require synaptic activity to be maintained. Pharmacological studies, summarized in Fig. 7 , suggest that the triggering of dedepression of NMDAR-mediated synaptic responses requires activation of NMDARs by HFS but not mGluRs or L-type Ca2+ channels. Furthermore, release of Ca2+ from internal stores is not required. Thus it is likely that like all other forms of NMDAR-dependent synaptic plasticity, this form of dedepression requires an NMDAR-dependent rise in postsynaptic calcium concentration.
Dedepression of AMPAR-mediated synaptic responses can also be generated (Mulkey and Malenka 1992) and has been suggested to require PKC (Daw et al. 2000) and PKA activity (Lee et al. 2000). To determine whether dedepression of NMDAR-mediated synaptic responses shared similar mechanisms, we examined the effects of inhibitors not only of these enzymes but also of a number of other protein kinases that have been implicated in LTP including CaMKII, Src, and MAPKs (summarized in Fig. 7). Surprisingly, only inhibitors of the MAPKs, ERK1/2, were effective in impairing the dedepression of NMDAR fEPSPs. It might be argued that the lack of effect of all other inhibitors may have been due to insufficient access of the drugs to their site of action. This seems unlikely because slices were both preincubated and perfused with concentrations of the drugs that have been reported to have clear effects in slice preparations. Moreover, the MAPK inhibitors were clearly effective using this same method of application. Loading CA1 pyramidal cells with an ERK1/2 inhibitor impaired dedepression of NMDAR EPSCs, suggesting that postsynaptic ERK1/2 activity is required.
It is important to note that some degree of dedepression remained when ERK1/2 inhibitors were applied either extracellularly (Fig. 4) or intracellularly by the patch pipettes (Fig. 6). There are two possible interpretations of this incomplete block of dedepression. If ERK1/2 activity was very strongly inhibited during these experiments, this result implies that ERK1/2 signaling alone is only partially responsible for the dedepression and that other signaling pathways contribute. However, it is also possible that the ERK1/2 inhibitors were not 100% effective and that if ERK1/2 activity had been completely blocked, the dedepression of NMDAR-mediated synaptic response also would have been completely prevented. Indeed, when loading cells postsynaptically with drugs or peptides, it is impossible to know what concentration of the reagent is reaching the critical synaptic site or how effective the reagent is at inhibiting the targeted enzymes. Despite these limitations to the interpretation of the experiments using ERK1/2 inhibitors, the results do strongly suggest that postsynaptic ERK1/2 activity importantly contributes to this form of dedepression.
It is unclear how HFS of NMDARs might lead to activation of MAPK cascades and the generation of dedepression. Calcium influx through NMDARs can activate the small GTPase Ras, which acts as a molecular switch to activate the MAPK cascade leading to ERK activation (Krapivinsky et al. 2003). Although MAPK activation influences gene expression the observation that dedepression manifests fairly rapidly after HFS (within 20 min) suggests it is unlikely that this effect (i.e., translocation of MAPK from dendrite to nucleus and subsequent trafficking of new gene product from the soma back to the dendrites) is responsible for mediating dedepression. Furthermore, the observation that the protein synthesis inhibitor cycloheximide had no effect on dedepression of NMDAR fEPSPs makes it unlikely that ERK1/2-dependent activation of translation (Banko et al. 2005; Giovannini et al. 2001; Kelleher et al. 2004) is required. A more attractive scenario is that MAPKs phosphorylate substrates that are targeted in or near dendrites. The findings that active ERKs are present in dendrites and phosphorylate dendritic voltage-dependent K+ channels support this notion (Yuan et al. 2002).
Activation of ERKs by Ras after NMDAR and CaMKII activation has been reported to drive synaptic delivery of AMPARs during LTP (Zhu et al. 2002). This raises the possibility that the dedepression of NMDAR-mediated synaptic responses may involve insertion of NMDARs into postsynaptic membranes. However, no change in synaptic NMDARs was observed after the manipulations that stimulated delivery of AMPARs to synapses (Zhu et al. 2002). Furthermore, LTD of NMDAR-mediated synaptic responses does not require endocytosis of NMDARs but rather appears to involve depolymerization of the actin cytoskeleton through a protein phosphatase 1–dependent mechanism (Morishita et al. 2005). Therefore we favor the hypothesis that activation of MAPKs during dedepression may promote actin polymerization in a manner that potentiates the depressed NMDAR-mediated synaptic responses. In this context, it is worth noting that an increase in f-actin bundles and filopodial density due to phosphorylation of spinophillin by ERK2 has been reported (Futter et al. 2005).
The results presented here provide strong evidence that, like synaptic AMPARs, synaptic NMDARs can be bidirectionally modified by different patterns of synaptic activity. However, several experimental findings make it clear that there are likely important differences in the mechanisms by which this occurs. First, LTP of synaptic AMPARs is normally reliable and robust, whereas, as discussed in the introduction, LTP of synaptic NMDARs is often difficult to generate. Indeed, only a small, nonsignificant potentiation of basal NMDAR fEPSPs was induced by the HFS in the absence of prior LTD (Fig. 1B). One possible explanation for these observations is that synapses may exist in different states (Montgomery and Madison 2004) such that the strength of AMPAR- and NMDAR-mediated transmission is at very different levels. Specifically, the difficulty in eliciting LTP of NMDAR-mediated synaptic responses combined with the relative ease of eliciting LTD and dedepression of these same responses may reflect the fact that most CA1 synapses in standard acute hippocampal slices express close to maximally potentiated NMDAR-mediated responses at the beginning of most experiments. Second, LTP and dedepression of NMDAR-mediated synaptic responses, when they do occur, often exhibit a time course distinctly slower than LTP and dedepression of AMPAR-mediated responses (Daw et al. 2000; Lee et al. 2000; Watt et al. 2004; Xiao et al. 1995). The mechanisms underlying the slower growth of NMDAR-mediated synaptic responses are unclear, one possibility being that the HFS simultaneously elicits a short-term depression of NMDAR-mediated responses. Third, as mentioned earlier, the mechanisms of LTD of NMDAR- and AMPAR-mediated synaptic responses differ (Morishita et al. 2005; Selig et al. 1995). Fourth, the results presented here indicate that the mechanisms of dedepression of NMDAR- and AMPAR-mediated responses also differ. Given that the role of different NMDAR subtypes in triggering LTD and LTP of AMPAR-mediated synaptic responses has been a contentious topic (e.g., Liu et al. 2004: Morishita et al. 2007), it will be important in future investigations to examine the role of specific subtypes of NMDARs in the bidirectional modulation of NMDAR-mediated synaptic responses.
In conclusion, we have demonstrated robust and reliable bidirectional regulation of NMDAR-mediated synaptic transmission by activity and provided evidence that this involves mechanisms distinct from those responsible for the activity-dependent bidirectional regulation of AMPAR-mediated responses. It is hoped that this initial analysis of the mechanisms underlying the dedepression of postsynaptic NMDARs will stimulate more work on this topic because the activity-dependent modulation of postsynaptic NMDARs will certainly have profound effects on their roles in a host of adaptive and pathological brain functions.
This work was supported by National Institute of Mental Health Grant 5 R37 MH-063394 to R. C. Malenka.
We thank members of the Malenka laboratory for helpful comments.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 by the American Physiological Society