Recent work has shown that some low-frequency stimulation (LFS) protocols can induce long-term potentiation (LTP) at hippocampal synapses. As LFS mimics certain aspects of low-frequency oscillations during slow-wave sleep, LFS-LTP may be relevant to processes of sleep-dependent consolidation. Here, alternating LFS (1 Hz) of heterosynaptic inputs arising in the medial septum and area CA3 induced LTP at hippocampal CA1 synapses of anesthetized rats. Remarkably, this LTP was absent when delivered 3 h, but not 8 or 24 h, after training in the hidden platform version of the Morris water maze, suggesting a time-specific occlusion of LFS-LTP following spatial learning. LTP assessed 3 h after training was intact in yoked swim controls and rats trained in darkness. Visible platform training resulted in heterogeneous effects, with about half of the animals showing LTP occlusion. Pharmacological experiments revealed that N-methyl-d-aspartate (NMDA)-receptor activation was required for both LFS-LTP and the retention of spatial learning. To test whether a learning-related, NMDA-dependent potentiation accounted for the occlusion effect, we blocked NMDA receptors immediately following spatial training. This manipulation reversed LTP occlusion 3 h after training. Together, these experiments indicate a mechanistic overlap between heterosynaptically induced LFS-LTP and processes mediating the consolidation of spatial information at hippocampal synapses.
Long-term potentiation (LTP) displays many properties known to govern learning and memory at the behavioral level (Bliss and Collingridge 1993; Martin and Morris 2002; Lynch 2004). Typically, LTP is induced by high-frequency stimulation (HFS) of afferent fibers (Bliss and Lomo 1973; Lynch 2004; Albensi et al. 2007), and there now is considerable evidence for an overlap of mechanisms mediating HFS-LTP and training-induced plasticity. Cued fear conditioning, motor training, and inhibitory avoidance training have all been shown to result in enhancements of synaptic strength in specific, task relevant forebrain areas such as the amygdala, motor cortex, and hippocampus (Rogan et al. 1997; Rioult-Pedotti et al. 2000; Monfils and Teskey 2004; Whitlock et al. 2006). Importantly, synaptic potentiation following behavioral training limits (occludes) the subsequent induction of LTP by electrical HFS (Rioult-Pedotti et al. 2000; Monfils and Teskey 2004; Whitlock et al. 2006), an observation that provides strong support for the notion that similar mechanisms mediate both learning and LTP induced by electrical HFS (Martin and Morris 2002).
To date, work linking LTP and training-induced synaptic enhancements has focused exclusively on LTP induced by HFS. Remarkably, however, some data suggest that low-frequency stimulation (LFS), typically used to elicit long-term depression (LTD) (Albensi et al. 2007), can also effectively induce LTP at some forebrain synapses (Thomas et al. 1996; Huang and Kandel 2007). For example, LFS (1 Hz) to area CA1 of the hippocampus produces a slow-onset form of potentiation at local synapses in vitro (Lanté, Cavalier, et al. 2006; Lanté, de Jésus Ferreira, et al. 2006). Furthermore, alternating (1 Hz) LFS of heterosynaptic inputs from the medial septum (MS) and CA3 area elicits long lasting (>4 h) LTP of field excitatory postsynaptic potentials (fEPSPs) in CA1 in vivo (Habib and Dringenberg 2009, 2010a). Importantly, LFS frequencies employed in these studies are in the range of endogenous slow oscillations present in the forebrain during natural slow-wave sleep (Steriade et al. 1993; Wolansky et al. 2006; Schall et al. 2008). Although considerable interest exists in the relation between slow, oscillatory brain states and sleep-dependent memory consolidation (Buzsáki 1998; Sirota and Buzsáki 2005; Born et al. 2006; Marshall et al. 2006), a direct link between hippocampal LFS-LTP and consolidation processes has not yet been investigated. Here, we address this issue by examining whether a preceding learning experience alters heterosynaptically induced LFS-LTP in the hippocampus, an approach equivalent to the “occlusion methodology” (Martin and Morris 2002) used to establish a mechanistic overlap between training-induced plasticity and HFS-LTP in several forebrain areas (Rioult-Pedotti et al. 2000; Monfils and Teskey 2004; Whitlock et al. 2006).
Materials and Methods
Experiments were performed on male Long-Evans rats (320–450 g) housed in a colony room under 12/12-h reversed light cycle. Procedures were conducted in accordance with guidelines of the Canadian Council on Animal Care and were approved by the Queen's University Animal Care Committee.
Water Maze Training
The water maze (a circular pool, 1.8-m diameter) was filled with water (23–25 °C) made opaque by adding nontoxic white paint. A circular escape platform (30 cm diameter, 2 cm below water surface, left stationary during training) was submerged in the center of 1 of the 4 pool quadrants. Rats were trained by administering 12 training trials, divided into 3 blocks of 4 trials each, with a 5-min rest period between blocks. For each trial block, 4 different start positions from the 4 cardinal compass points were used in random order. For each trial, the animal was released into the pool facing the pool wall and was then allowed to swim freely, for a maximum of 60 s, to locate the hidden escape platform. After mounting the platform, the animal was permitted to remain there for 15 s before commencing the next trial. Animals that failed to locate the platform within 60 s were manually guided to, and placed onto, the platform. The latencies to reach the platform were recorded by an experimenter. One group of rats was trained on the visible (cued) version of the task, where the platform location was indicated by a visible ensign (10-cm high plastic cylinder marked with black stripes and topped with a blue rubber ball). To ensure that rats used the local cue as a guide during training, the cued platform was randomly moved to a new pool quadrant for each trial (each quadrant used once per trial block). Further, for a group of yoked controls, one rat was matched to each animal in the hidden platform group for the swim duration for each trial, but swimming occurred without a platform in the pool. Finally, one group of rats was trained in complete darkness by using the “Nightshot” function of a digital video camera (Sony Super SteadyShot DCR-TRV17, Sony Corporation of Canada).
For the pharmacological experiments, groups of rats received either the N-methyl-d-aspartate (NMDA)-receptor antagonist (6)-3-(2-carboxypiperazin-4-yl) propyl-1-phosphonic acid (CPP; 10 mg/kg, intraperitoneal [i.p.]; Sigma, Oakville, Ontario) or saline immediately after hidden platform training. Some of these rats were given another 4 test trials (same platform location; all trials video-recorded for subsequent, blind analyses) 3 h after the end of training, with the first of these trials serving as a retention test, while the next 3 trials were used as an index of reacquisition. Only those rats that met a predetermined acquisition criterion for the initial 12 training trials (escape latency of >20 s on the last training trial) were included in the subsequent retention testing.
Electrophysiological experiments (under urethane anesthesia, 1.5 g/kg, i.p., administered as 3 × 0.5 g/kg every 20 min, supplemented as necessary; additional marcaine [5 mg/mL] applied to the skin overlying the skull; body temperature held at 36–37 °C) were carried out so that LFS was delivered exactly 3, 8, or 24 h after the end of water maze training (the first urethane dose was given 2 h prior to LFS). Small skull holes were drilled and electrodes lowered into the CA3 region (AP −4.16, ML +3.0, V −4.0), the contralateral CA1 area (AP −4.16, ML −3.0 V −3.0), and the MS (AP 0.0, ML 0.0, V – 6.5).
Stimulation of CA3 (0.2-ms pulses every 30 s) was provided by a concentric bipolar electrode (Rhodes Medical Instruments Series 100, David Kopf, Tujunga, CA, USA) connected to a stimulus isolator providing constant-current output (PowerLab/16 s system with ML 180 Stimulus Isolator, ADInstruments, Toronto, Ontario, Canada). MS stimulation was provided through a monopolar electrode (Teflon-insulated stainless steel, 125-μm diameter) delivering negative, constant-current pulses (0.2 ms duration, 0.5 mA). Recordings of fEPSPs in CA1 were obtained using a Teflon-insulated stainless steel electrode (125-μm diameter), followed by amplification, filtering (0.3–1 kHz), and digitization (10 kHz) using the PowerLab system running Scope software (v. 4.0.2).
For each animal, input–output curves were established by recording fEPSPs elicited by CA3 stimulation (0.1–1.0 mA, 0.1 mA increments); the stimulation intensity eliciting 50–60% of the maximal fEPSP amplitude was used for further data collection. Subsequently, 60 baseline fEPSPs (every 30 s) were recorded, followed by delivery of the LFS protocol, consisting of 50 single pulses to each the MS and CA3 area (0.5 Hz for each), alternating between the 2 sites (1000-ms interstimulus interval; see Habib and Dringenberg 2009, 2010b). Recordings of fEPSPs (every 30 s) continued for 2 h after LFS, followed by perfusion (10% formalin) of the rat and brain extraction. Standard histological techniques were used to verify electrode placements. Data obtained with inaccurate placements were excluded from the data analysis.
Data are expressed as mean ± standard error of mean (SEM). The maximal fEPSP amplitude was analyzed offline by the Scope software. Subsequently, amplitude data were averaged over 10-min intervals, and these averages were normalized by dividing all data for each rat by the average baseline (pre-MS-CA3 alternating stimulation) amplitude of that animal. Water maze and electrophysiological data were analyzed using analyses of variance (ANOVA) and, where statistically appropriate, simple effects tests using the software package CLR ANOVA (v.1.1, Clear Lake Research, Inc., Houston, TX, USA).
Spatial Training Occludes LFS-LTP
In order to assess whether water maze training altered the induction of LFS-LTP in area CA1, independent groups of rats were trained on either the hidden (n = 28) or visible (cued; n = 7) platform version of the task. Both groups showed significant decreases in escape latency over 12 training trials (F11,275 = 15.1, P < 0.001), indicative of successful task acquisition (Fig. 1). Further, rats trained with a visible platform displayed shorter latencies during the initial training trials (Fig. 1; group by trial interaction, F11, 275 = 2.3, P < 0.008).
Electrophysiological procedures under urethane anesthesia were carried out so that the induction of LFS-LTP occurred at 3, 8, or 24 h following the end of behavioral training. LFS-LTP in the hippocampal CA1 area was induced by applying a 1 Hz stimulation protocol, with successive stimulation pulses alternating between the MS and contralateral CA3 field, as described previously (Habib and Dringenberg 2009, 2010a). This LFS protocol successfully induced LTP when applied to rats that had received hidden platform training 8 h (n = 6) or 24 h (n = 7) prior to LTP induction, as indicated by significant enhancements of fEPSP amplitude in CA1 to 119% and 117% of baseline, respectively (Fig. 2A; values are averaged over the last 30 min of the experiment; effects of time, 8 h group: F14,70 = 3.7, P < 0.001; 24 h group: F14,84 = 7.4, P < 0.001). These levels of LTP are similar to those reported previously for task-naïve rats (Habib and Dringenberg 2009, 2010a). In contrast, rats (n = 7) that received LFS at 3 h following spatial training failed to exhibit synaptic potentiation, with fEPSP amplitude at 98% of baseline during the last 30 min of the experiment (Fig. 2A; effect of time, F14,84 = 0.6, P = 0.85). Accordingly, an ANOVA comparing the 3 groups (3, 8, and 24 h) yielded a significant group by time interaction, F28,238 = 2.6, P < 0.001, indicative of lower fEPSP amplitude towards the end of the experiment in rats tested at 3 h after training (follow-up comparisons yielded significant time by group effects when the 3 h group was compared against each of the other conditions, P < 0.001).
Next, we assessed whether the apparent occlusion of LFS-LTP at 3 h after training is selective for spatial aspects of water maze training. One group of rats (n = 7) served as yoked swim controls, consisting of animals individually matched to spatially trained rats for swim time (per trial) in the absence of an escape platform. In these rats, LFS at 3 h after swimming resulted in a clear enhancement of fEPSP amplitude (Fig. 2B; average amplitude at 119% of baseline; F14,84 = 3.6, P < 0.001). Thus, stress and other factors associated with swimming did not account for the LTP occlusion observed 3 h after spatial training. A further group of rats (n = 7) received LFS 3 h after training in the visible platform version of the task. As a group, these rats also showed fEPSP enhancement (Fig. 2B; 115% of baseline; F14,84 = 1.9, P = 0.039), but this effect was quite variable, as 4 of 7 rats showed LTP, whereas the rest did not. An ANOVA indicated a significant difference in fEPSP amplitude between spatially trained rats and swim controls (group by time interaction, F14,168 = 2.8, P = 0.001), but not between rats that received spatial and cued water maze training (P = 0.14).
It is possible that the heterogeneous effect on LTP in rats undergoing visible platform training was due to the fact that animals experienced and stored spatial information, even when a local cue was available to indicate the platform location (see Discussion section). To address this possibility, another group of rats (n = 7) received water maze training in complete darkness to rule out the use of spatial cues. Under these conditions, task acquisition was significantly impaired relative to animals trained under normal illumination (Fig. 3A; group by trial interaction, F11,363 = 3.6, P < 0.001; group effect also significant, P < 0.001). Further, rats trained under darkness showed no sign of LTP occlusion (Fig. 3B), with all rats exhibiting clear LTP when the LFS protocol was delivered 3 h after training (129% of baseline; ANOVA comparing rats trained under light and darkness, group by time interaction, F14,154 = 4.7, P < 0.001). Thus, it appears that the elimination of spatial cues during training allowed for the expression of LFS-LTP.
The occlusion of LFS-LTP suggests the occurrence of an endogenous form of synaptic potentiation 3 h after hidden platform training. Consequently, we plotted fEPSP amplitudes obtained during input–output curves (Fig. 4) for rats showing LTP occlusion (3 h after hidden platform training, n = 7) against all other experimental conditions (i.e., groups that showed LTP, n = 27). At low stimulation intensities (0.1–0.3 mA), rats showing occlusion displayed a trend toward larger fEPSP amplitudes relative to rats with intact LTP, but this difference did not reach statistical significance (Fig. 4; P's > 0.05).
Consolidation of Spatial Learning Involves NMDA Receptors
If an endogenous form of LFS-LTP contributes to the consolidation and/or retention of spatial information at 3 h after training, then rats should remember the location of the hidden platform at this time interval. Also, prevention of endogenous LFS-LTP by NMDA-receptor antagonist treatment (Habib and Dringenberg 2009) should interfere with consolidation or retention of spatial information at this time point.
We tested these hypotheses by training rats (n = 24) with a hidden platform (same procedure as described), followed by a retention test 3 h after training. Rats that successfully acquired the task showed a decrease in escape latencies over the 12 acquisition trials (Fig. 5). Immediately after training, one group (n = 11) was administered the NMDA-receptor antagonist CPP (10 mg/kg, i.p.), while another group (n = 13) received saline (1 mL/kg, i.p.). Retention testing was carried out 3 h after training. Four trials (platform in the same location as during training) were administered, with the first trial serving as a “retention” test, while the subsequent 3 trials were taken as a measure of “reacquisition.” This test revealed that on the first (retention) trial, rats that had received CPP following training exhibited longer search latencies to find the platform relative to saline-treated animals (Fig. 5; 20.6 ± 4.5 and 9.2 ± 2.5 s, respectively). However, over the 3 subsequent reacquisition trials, performance in the CPP rats improved to levels similar to those in saline-treated rats (Fig. 5; third reacquisition trial latencies of 10.1 ± 1.5 and 9.1 ± 1.3 s for CPP and saline rats, respectively).
ANOVA for latencies on the retention and reacquisition trials confirmed an overall significant group effect (F1,22 = 9.3, P = 0.006), with trial-by-trial analyses revealing a group difference on the “retention” trial (F1,22 = 5.4, P = 0.03), but not on any of the 3 reacquisition trials (P's > 0.15). Together, these experiments suggest that NMDA receptors participate in the stabilization of spatial information over the same time interval as the occlusion effect seen for LFS-LTP.
Reversal of LFS-LTP Occlusion by NMDA-Receptor Blockade After Spatial Training
The form of LFS-LTP induced by alternating MS-CA3 stimulation requires NMDA-receptor activation (Habib and Dringenberg 2009). If spatial training induces an endogenous form of LTP involving the similar mechanisms, blockade of the training-induced LTP should reverse the occlusion normally seen 3 h after spatial training, thus rescuing LFS-LTP. To examine this hypothesis, we tested a group of rats (n = 8) that received the NMDA-receptor antagonist CPP (10 mg/kg, i.p.) immediately after completion of hidden platform training (as above). Remarkably, in these rats, LFS delivered 3 h after training resulted in LTP, with fEPSP amplitude reaching 116% of baseline during the last 30 min of recording (Fig. 6). In contrast, untrained rats (n = 7) given CPP 1 h prior to MS-CA3 stimulation failed to show LTP and, in fact, exhibited a depression-like effect (Fig. 6; amplitude at 81% of baseline). These data confirm the role of acute NMDA-receptor activation in LFS-LTP and reveal that interference with NMDA-receptor functioning after training reverses LFS-LTP occlusion, most likely by preventing an endogenous form of LTP elicited by spatial training.
An ANOVA comparing groups receiving post-training CPP, acute CPP, and untreated, spatially trained rats (same group as Fig. 2 showing occlusion) revealed a significant group by time interaction, F28,266 = 4.3, P < 0.01. Further analyses indicated a significant group by time interaction for comparisons of the post-training CPP and training (no drug) groups, F14,154 = 1.9, P = 0.024, and the training (no drug) and acute CPP groups, F14,182 = 2.4, P < 0.004.
Slow-wave sleep in mammals contains a pronounced, slow rhythm at a frequency of about 1 Hz, which has been hypothesized as a potential mechanism mediating slow-wave sleep-dependent memory consolidation in hippocampal and neocortical networks (Steriade et al. 1993; Buzsáki 1998; Sirota and Buzsáki 2005; Born et al. 2006; Wolansky et al. 2006). Consistent with this notion, electrical stimulation protocols employing 1 Hz stimulation trains can be effective in eliciting synaptic potentiation at hippocampal and amygdaloid synapses (Huang and Kandel 2007; Lanté, Cavalier, et al. 2006; Lanté, de Jésus Ferreira, et al. 2006; Habib and Dringenberg 2009, 2010a). Here, we studied the effect of water maze training on the subsequent induction of a form of heterosynaptic LFS-LTP in CA1 induced by alternating, 1 Hz stimulation of the hippocampal CA3 field and fibers originating in the MS (Habib and Dringenberg 2009, 2010a). Delivery of LFS at 8 or 24 h after training on the hidden platform version of the task resulted in an enhancement of fEPSPs in CA1 to levels equivalent to those seen in the absence of training (Habib and Dringenberg 2009, 2010a). However, rats that received the LFS protocol 3 h after spatial training did not show significant fEPSP potentiation, whereas yoked control rats (that swam for the same duration per trial as each rat in the hidden platform group) and rats trained in darkness (to eliminate spatial cues) showed intact LTP. Thus, stress and other factors associated with swimming in the maze do not account for the LTP occlusion seen 3 h after hidden platform training. Further experiments confirmed that rats remember the location of a hidden platform 3 h after training and that NMDA receptors participate both in the consolidation process over this time interval and the induction of LFS-LTP. Finally, NMDA-receptor blockade immediately following spatial training reversed the occlusion effect 3 h later, presumably by interfering with the induction of an endogenous potentiation process. Together, these results provide strong support for the occurrence of an activity-dependent synaptic potentiation, present at a specific time window following training, which shares mechanisms and therefore competes with LFS-induced LTP in the CA1 area in vivo.
Occlusion of electrically induced LTP by preceding training constitutes one of the fundamental criteria to demonstrate a role for LTP-like processes in memory formation (Martin and Morris 2002). Motor and passive avoidance training results in an enhancement of fEPSPs in the primary motor cortex and hippocampus, respectively, leading to a reduction (occlusion) of HFS-LTP (Rioult-Pedotti et al. 2000; Monfils and Teskey 2004; Whitlock et al. 2006). The occlusion of LFS-LTP noted here 3 h after spatial training is consistent with the induction of an endogenous form of LTP. Training-induced LTP should be evident by an increase in fEPSP amplitude prior to electrical LTP induction (i.e., baseline fEPSPs). However, our analysis of the initial input–output curves failed to reveal a significant enhancement in baseline fEPSP amplitude, even though it is noteworthy that fEPSPs were consistently larger for spatially trained rats at lower stimulation intensities (0.1–0.3 mA). It is likely that training-induced synaptic potentiation is a spatially distributed phenomenon. Consequently, for future work, the use of multielectrode arrays should provide a more sensitive assay to detect local changes in synaptic efficacy throughout the complex CA1 network (see Whitlock et al. 2006).
In order to determine whether training-induced LTP occlusion was specific to the hidden platform version of the task, we delivered LFS to rats trained on the visible platform version of the task, as well as rats that were yoked to hidden platform rats for swim time without a platform present in the maze. Yoked controls showed intact LTP, while rats trained with a visible platform exhibited a slight, nonsignificant reduction in LTP, which was intermediate to potentiation levels in yoked rats and rats tested 3 h after spatial training. Further analyses revealed that LTP in the visible platform group was highly heterogeneous, with 3 and 4 animals demonstrating occlusion and intact LTP, respectively. These results may be due to the fact that rats can solve the visible platform version of the water maze by adopting either a spatial or nonspatial navigational strategy, with many individual animals consistently preferring one of these strategies (McDonald and White 1993, 1994). Thus, even during visible platform training, rats are exposed to, and may acquire, hippocampal-dependent spatial information, which would result in LTP occlusion in a subset of animals.
In order to rule out the exposure to, and use of visible, spatial cues, we tested a further group of rats that received water maze training in complete darkness. As expected, task acquisition was significantly impaired under these conditions, even though it is noteworthy that rats did exhibit some improvement over the 12 training trials, largely because rats learned to avoid the pool wall during swimming and searching for the platform. Importantly, dark training resulted in the full expression of LTP in all rats tested under these conditions. Thus, training in the absence of visible, spatial cues did not lead to LTP occlusion, suggesting that the exposure to spatial information is an important requirement for the occlusion effect to occur.
The occlusion of LFS-LTP exhibited a clear, temporal specificity, with occlusion seen at 3, but not 8 or 24 h after hidden platform training. Thus, spatial learning may induce an endogenous form of synaptic potentiation during the earlier stages of memory consolidation in local hippocampal circuits, thought to precede systems-wide consolidation (occurring over days, weeks, or years) involving multiple brain areas (Martin and Clark 2007). Activation of NMDA receptors provides a cellular mechanism underlying the stabilization of memory after an initial encoding event (Lynch 2004; Abraham and Williams 2008). For example, hippocampal NMDA-receptor blockade using CPP immediately prior to water maze training allows for the rapid acquisition of an escape response, but impairs retention at 24 h, indicative of a contribution of NMDA receptors to post-training consolidation (Holahan et al. 2005; McDonald et al. 2005), even though acquisition impairments following NMDA-receptor blockade have also been reported (Morris et al. 1986). Here, post-training administration of CPP impaired retention performance 3 h after training, further supporting the role of NMDA receptors in memory consolidation. The impairment on the retention trial could not be accounted for by long-lasting side-effects of CPP (e.g., reduced swim speed and impaired visual acuity), since drug-treated rats rapidly re-learned the task to control levels over the 3 reacquisition trials. The observations that, like consolidation, LFS-LTP requires NMDA-receptor activation and exhibits a relatively slow, gradual development over about 2 h after induction (present experiments; Habib and Dringenberg 2009) also support its potential role as a temporally specific mechanism mediating the consolidation of spatial information.
Based on these considerations, we reasoned that interference with a training-induced, NMDA-receptor-dependent consolidation process might reverse the occlusion of LFS-LTP after spatial training. This hypothesis was confirmed in rats treated with CPP immediately after hidden platform training, a treatment that reversed occlusion (in addition to impairing consolidation; see above) and allowed for the successful induction of LTP in spatially trained rats. These observations provide further evidence for the notion that heterosynaptic LFS-LTP may provide a mechanism mediating consolidation of hippocampal-dependent learning experiences.
We conducted the electrophysiological experiments under urethane anesthesia, while previous work showing HFS-LTP occlusion was conducted either in vitro (Rioult-Pedotti et al. 2000) or in freely moving rats (Monfils and Teskey 2004; Whitlock et al. 2006). Urethane offers advantages over other preparations, including the presence of well-preserved patterns of spontaneous, oscillatory activity in the neocortex and hippocampus, which include the slow (∼1 Hz) oscillation, a hallmark characteristic of human and rodent slow-wave sleep (Wolansky et al. 2006). In fact, cyclic alternations between deactivated and activated brain states under urethane closely resemble those in naturally sleeping rats (Clement et al. 2008). This mimicry of natural sleep by urethane might create conditions favoring the expression of sleep-related consolidation processes.
It is also important to acknowledge that changes in brain states (i.e., hippocampal slow-wave activity) are known to influence fEPSPs in the CA1 area (Schall et al. 2008). Thus, it is conceivable that training-related alterations in hippocampal slow waves may contribute to the differences (occlusion vs. intact) in LTP observed in the various experimental groups. Thus, continuous monitoring of slow-wave activity and brain state throughout the experiment will be an important addition for future work on the LTP phenomenon described here. However, the nonsignificant changes in fEPSPs during input–output curve recordings (see above) do not suggest that significant differences in brain state were present among spatially trained and control animals, at least during this phase of the experiment. In fact, the reversal of the occlusion effect by NMDA-receptor blockade immediate following spatial training suggests that potential brain state changes and related plasticity/consolidation mechanisms were present during an early, post-training consolidation period, rather than the time of LTP induction (at which time the NMDA antagonist was no longer active, based on the successful induction of LTP). Future work employing chronic recoding methods to continuously monitor spontaneous slow-wave activity during the post-training period is required to critically assess this hypothesis.
The findings reported here may also be relevant to the concept of “metaplasticity”, first introduced by Bienenstock, Cooper, and Munro (BCM theory) in 1982. According to their hypothesis, the ability of a neuron to express either LTP or LTD in response to a given input is not static. In fact, the threshold for LTD versus LTP induction varies systematically with the level of previous, postsynaptic activity of a given neuron, with higher previous activity raising the threshold for LTP induction (and facilitating LTD), while reduced activity exerts the opposite effect (Bienenstock et al. 1982; Abraham 2008; Cooper and Bear 2012). The BCM model provides a complementary or alternative explanation for occlusion effects, since it predicts that high levels of synaptic activity during (or after) learning raises the threshold for LTP induction and facilitates LTD, a prediction consistent with previous work measuring HFS-LTP (Rioult-Pedotti et al. 2000; Monfils and Teskey 2004; Whitlock et al. 2006). The occlusion effect seen in the present study suggests that similar, metaplasticity principles may also apply to forms of LTP elicited by LFS protocols. Also, it is of interest to note that metaplastic changes in the LTD/LTP threshold lack the input specificity normally associated with LTP, that is, the threshold shift appears to occur heterosynaptically at many or all excitatory synapses on a given neuron (Abraham et al. 2001). This lack of input specificity may account for the fact that we saw LTP occlusion, but only a modest (nonsignificant) change in fEPSPs during the input–output curves. Presumably, learning resulted in localized, distributed increases in synaptic strength, which are harder to detect than a more generalized increase in LTP threshold due to elevated hippocampal activity during and/or after the training experience.
In summary, the field of memory consolidation has raised fundamental questions regarding mechanisms by which sleep and sleep-related oscillatory brain activity can influence plasticity induction and maintenance at central synapses (Buzsáki 1998; Sirota and Buzsáki 2005; Born et al. 2006; Habib and Dringenberg 2010b). The present dataset provides the first evidence for a mechanistic overlap between a form of heterosynaptic LFS-LTP in area CA1 and endogenous plasticity mechanisms that might mediate earlier stages of memory consolidation in hippocampal networks.
This work was supported by the National Sciences and Engineering Research Council of Canada (NSERC Grant 203175, to H.C.D.).
We thank Deanna Choi for assistance with the behavioral analyses. Conflict of Interest: None declared.