Report

Abolition of Long-Term Stability of New Hippocampal Place Cell Maps by NMDA Receptor Blockade

See allHide authors and affiliations

Science  26 Jun 1998:
Vol. 280, Issue 5372, pp. 2121-2126
DOI: 10.1126/science.280.5372.2121

Abstract

Hippocampal pyramidal cells are called place cells because each cell tends to fire only when the animal is in a particular part of the environment—the cell's firing field. Acute pharmacological blockade of N-methyl-d-aspartate (NMDA) glutamate receptors was used to investigate how NMDA-based synaptic plasticity participates in the formation and maintenance of the firing fields. The results suggest that the formation and short-term stability of firing fields in a new environment involve plasticity that is independent of NMDA receptor activation. By contrast, the long-term stabilization of newly established firing fields required normal NMDA receptor function and, therefore, may be related to other NMDA-dependent processes such as long-term potentiation and spatial learning.

The ability of rodents to learn and remember features of a new environment is thought to require the formation in the animal's brain of a cognitive map—a neural representation of space. In 1971, O'Keefe and Dostrovsky (1) proposed that a rat's position in space is encoded by the coordinated activity of individual hippocampal pyramidal cells [place cells, recently reviewed by (2)]. Such encoding is possible because each place cell tends to fire only when the rat (or mouse) is in a cell-specific part of the current environment, the cell's “firing field.” The conjoint activity of place cells is therefore thought to be the basis of a map of the environment that the animal uses for solving spatial problems. In this sense, the cognitive map serves as a cellular substrate for spatial memory (3). Place cells have two other properties that make them attractive as elements of a spatial memory system. The first is environmental stability—a given cell has the same firing field in each of many exposures to the same environment, for as long as the cell is identifiable (up to 6 months) (4). The other is environmental specificity—the firing field of a place cell in one environment does not predict its field in a second, distinct environment (5, 6). Thus, when an animal is put into a new environment, each pyramidal cell changes its positional firing pattern in an entirely unpredictable fashion: The field of any cell can change in firing rate, shape, or position (or a combination of all three) or turn off or on, irrespective of what other cells do. This process is called “remapping” and reflects the formation of a new hippocampal map for the novel environment. Once formed, this new map is also stable and does not interfere with existing maps of familiar environments (7,8).

That rodents can rapidly form stable representations of new environments raises the following question: What are the cellular mechanisms whereby firing fields are first formed and once formed are then maintained? One candidate mechanism is long-term potentiation (LTP) or, more precisely, the plastic processes that underlie LTP. LTP is a long-lasting, activity-dependent enhancement of synaptic strength that has been extensively studied in the hippocampus (9). One type of LTP that appears to be important for spatial memory is the NMDA receptor–dependent form that occurs at the Schaffer collateral pathway that connects pyramidal cells of the CA3 region to those of the CA1 region. Typically, pharmacological or genetic disruption of this type of LTP results in impaired performance in tasks that require spatial memory (10). Thus, NMDA receptors may play an essential role in the formation and maintenance of spatial maps. To test this idea, we used the competitive NMDA receptor antagonist CPP [(±)-3-(2-carboxypiperazin-4-yl)propyl-1- phosphonic acid; RB, Natick, Massachusetts] and addressed three questions: (i) Does acute blockade of NMDA receptors throughout the brain produce a degradation of the positional firing patterns of CA1 place cells in a familiar environment? (ii) Does acute blockade of NMDA receptors prevent remapping when the rat is put into a new environment? (iii) Finally, does this blockade affect the short- or long-term stability of newly formed place fields?

Our experimental strategy, summarized in Fig. 1, was based on the stability of place cells in a familiar environment and the development of new fields by the same cells in a novel environment (6, 7). We used a 76-cm-diameter gray cylinder with a white cue card as the familiar environment and a geometrically identical white cylinder with a black cue card as the novel environment. Animals were injected with either CPP or saline before their first exposure to the novel environment. Examples of positional firing patterns in the two environments of four pyramidal cells simultaneously recorded from a saline-injected rat (Fig. 2, A and B) and four pyramidal cells simultaneously recorded from a CPP-injected rat (Fig. 2, C and D) are shown in Fig. 2 (11). Each row shows firing rate maps for a single cell during 10 recording sessions over 2 days. The maps are grouped first according to the recording apparatus (gray cylinder on the left; white cylinder on the right) and then by the time order of the session.

Figure 1

Experimental protocol. Recordings were made during three experimental days called D0, D1, and D2. On the first day (D0), at least two well-isolated place cells were identified, and three sessions were run in the familiar gray cylinder. The first session (D0G1) was used to characterize the cell set, the second (D0G2) to test the effects of rotating the cue card, and the third (D0G3) to test whether the card rotation effects were reversible. For all rats, we found that rotating the white cue card 90° caused equal rotations of each firing field and that the field rotations were reversible, establishing that visual stimuli controlled firing fields. At the start of the next day (D1), a gray cylinder session (D1G0) was run to see if the cell recordings were stable. If so, the rat was injected with either physiological saline or CPP (10 mg/kg) and returned to its home cage for 1 hour. Three sessions were then run in the familiar gray cylinder, the novel white cylinder, and the familiar cylinder (D1G1, D1W1, and D1G2, respectively). Session D1G1 allowed us to ask if the drug affected established place cells, session D1W1 allowed us to see if CPP interfered with remapping, and session D1G2 allowed us to tell if exposure to the new environment disrupted the established firing fields. These sessions took a total of about 1 hour, after which the rat was returned to its home cage for 1 hour. Two more sessions were then run in the white cylinder and the gray cylinder. Session D1W2 allowed us to ask if the remapping in the novel environment was stable, and session D1G3 provided both a check of cell stability and a baseline for day 2 recordings. On day 2, after the drug had ceased to act (seeFig. 4), recording sessions were divided into two pairs separated by 1 hour in the home cage. Session D2G1 allowed us to check for cell stability in the familiar environment, and session D2W1 allowed us to see if the remapping on the first day was stable. Session D2G2 provided yet another check of cell stability, and session D2W2 allowed us to see if the firing fields in the white cylinder were the same as in session D2W1.

Figure 2

Examples of the firing fields of four cells from a saline-injected rat and four cells from a CPP-injected rat for all sessions on days 1 and 2 of the protocol. Each square pixel in a rate map represents a 3.5 cm by 3.5 cm area in the apparatus. Yellow encodes regions that the rat visited in which the cell never fired. Orange, red, green, blue, and purple pixels encode progressively higher firing rates and are autoscaled in each session. A gray pixel signifies the field center, and white pixels were not visited. See (6, 7) for details. (A and B) Saline-injected rat in familiar (A) and novel (B) environments. Firing rate maps for each pyramidal cell are shown as a row. The rate maps are sorted first according to the recording chamber as indicated by the outlines around each map and then according to time order. The gray backgrounds highlight all day 1 postinjection sessions. (C and D) CPP-injected rat in familiar (C) and novel (D) environments . Firing rate maps for four pyramidal cells recorded for the 10-session protocol spanning 2 days. Cell 4 was lost before the last two sessions (D2G2 and D2W2).

These maps illustrate four basic findings. First, blocking NMDA receptors did not interfere with a previously formed map. In the familiar gray cylinder, each cell had the same firing pattern after the injection as it did before (sessions D1G0 and D1G1), showing that place cells are as stable in a familiar environment after injections of CPP as they are after injection of saline. Second, blocking NMDA receptors did not prevent remapping in a novel environment. During the first exposure of the rats to the novel white cylinder (D1W1), the firing patterns of the place cells did not obviously resemble those in the gray cylinder (D1G1). Third, despite the blockade of NMDA receptors, the remapping seen in the novel white cylinder during the first session on day 1 persisted for at least 1.5 hours until the second session in the white cylinder (D1W2). Fourth, the most profound effect of blocking NMDA receptors was to abolish the long-term stability of the map in the novel environment. In the saline-injected rat, the remapping established on day 1 (D1W1 or D1W2) remained stable on day 2 (D2W1 and D2W2), but in the CPP-injected rat the new map formed on day 1 was replaced on day 2 with another new map. Thus, CPP prevented the first remapping from being stabilized.

To quantify these results, we compared positional firing patterns in pairs of sessions to look for stability or remapping before, during, and after NMDA channel blockade. We calculated a “similarity” score for a session pair by computing the correlation between the firing rates on a pixel-by-pixel basis (12). The mean similarity for the cells of each rat was computed and then averaged across rats to get group means (Fig. 3). The high similarity for preinjection and postinjection sessions in the gray cylinder (D1G0/D1G1) indicates that established firing fields are not significantly affected by CPP. This persistence is in agreement with studies showing that NMDA blockade does not interfere with established LTP in slices or with previously formed spatial memories in intact animals (9, 10). Moreover, CPP did not degrade the firing properties of individual place cells. We measured several properties of firing fields and found no significant effect of drug on field size, peak firing rate, coherence, information content, or signal-to-noise ratio. CPP also did not have any significant effect on average running speed in the familiar environment (13).

Figure 3

Comparisons of firing pattern similarity in selected pairs of sessions. Each comparison is shown as two bars indicating the mean similarity score (±SEM) for saline-injected rats (red) and CPP-injected rats (blue). Above the label for each comparison is the probability that the mean similarity for saline- and CPP-injected rats is equal by t tests, corrected for the number of tests. **, P < 0.001.

When the same cells were recorded during the animal's first introduction to the novel white cylinder, a remapping was seen for all saline- (6 of 6) and most CPP- (5 of 6) injected rats (comparison D1G1/D1W1, Fig. 3). For both saline- and CPP-injected rats, the first session in the gray cylinder (D1G1) was significantly more similar to later gray cylinder sessions than to the first session in the white cylinder (D1W1) (14). Furthermore, in all but the one CPP animal that did not remap, the firing fields of simultaneously recorded cells in the novel environment changed independently of each other. Comparing the first two white cylinder sessions (D1W1/D1W2) indicates that the remapping was stable for at least 1.5 hours for both saline- and CPP-injected rats. However, the initial remapping in rats injected with CPP tended to be less complete than in saline rats. By inspecting rate maps, we saw that the firing pattern in the novel environment partially resembled the pattern in the familiar environment for one or more cells in each rat. This residual discharge was absent in the second white cylinder session (D1G1/D1W2) except in the one rat injected with CPP that did not remap.

Although CPP had only relatively minor effects on remapping during day 1, it abolished the long-term stability of the newly formed map: A second remapping occurred for each CPP rat on day 2, as if they had not previously seen the white cylinder (D1W2/D2W1, mean similarity = 0.03 ± 0.02). The second remapping also did not resemble the original gray cylinder map (D2G1/D2W1, mean similarity = 0.01 ± 0.02). By contrast, the day 1 remapping was stable on day 2 in all saline-injected rats [D1W2/D2W1, mean similarity = 0.47 ± 0.05, t(10) = 7.08,P < 0.001 compared with CPP]. For CPP-injected rats, the new firing patterns in the white cylinder on day 2 (after the drug's effects had worn off) (15) were stable for at least 1.5 hours (D2W1/D2W2). The firing fields in the gray cylinder persisted from the first to the last session (familiar environment, D1G1/D2G2) in both groups, indicating that the recordings were stable for the duration of the experiment.

One possible explanation of the second remapping is that CPP acts as a discriminative stimulus for state-dependent learning (16). In this state-dependent view, the combination of CPP and the novel cylinder on day 1 is effectively a different environment than the novel cylinder on day 2, when CPP has ceased to act. According to this explanation, however, the combination of CPP and the familiar cylinder should cause a remapping from the predrug map in the familiar cylinder, which did not occur. In addition, we made a second injection of CPP in one rat on day 2 after the two white cylinder sessions. The firing fields stayed in the day 2 patterns, suggesting that the remapping on day 2 was not due to state-dependent learning but rather was due to instability of the day 1 map.

To determine how effectively the 10-mg/kg dose of CPP blocks NMDA receptors, we examined primed-burst potentiation (17) in awake, freely moving rats (Fig. 4) (18). Primed-burst potentiation is an activity-dependent enhancement of synaptic strength that is similar to LTP in its dependence on NMDA receptor activation but is of shorter duration, so that it can be tested repeatedly without saturation. A primed burst with no drug on day 0 caused robust potentiation of the population spike. By contrast, there was no potentiation on day 1 when primed bursts were delivered 1.5 and 3 hours after CPP injection, showing effective blockade of NMDA receptors. On day 2, a primed burst again caused potentiation, showing the drug had ceased to act. There was a significant effect of drug condition [F(3,12) = 7.40,P < 0.01] and a significant condition × time interaction [F(18,72) = 2.60, P < 0.01] in a two-way analysis of variance comparing the four conditions (before CPP injection and 90 min, 180 min, and 24 hours after injection). Subsequent analysis showed that the 90- and 180-min conditions were significantly different from baseline and 24 hours. Thus, the dose of CPP used for place cell recordings effectively blocked NMDA receptors at the time of day 1 recordings and ceased to act during day 2 recordings.

Figure 4

CPP reversibly blocks primed-burst potentiation (PBP). (A) Representative records of the population spike before (black) and 5 min after (red) primed-burst stimulation in the absence of CPP. (B) The average amplitude of the population spike in the CA1 region of the hippocampus, normalized to the average baseline value before primed-burst stimulation (PBS) (N = 5, mean baseline = 1.33 ± 0.12 mV, not significantly different among the four conditions). Vertical bars, SEM; *, P < 0.05 compared with baseline.

In a previous study investigating the role of NMDA receptors in spatial mapping, McHugh et al. (19) recorded from mice with a selective knockout of the NMDA receptor in pyramidal cells of the CA1 region. They found that place cells in the CA1 region of the knockout mice had firing fields that were somewhat abnormal but were stable for at least 1 hour. McHugh et al. (19) attributed the inability of NMDA receptor subunit 1 knockout mice to solve spatial problems to a defect in a higher order property of place cells: that cells with overlapping fields in the knockout mice do not tend to fire at the same time and therefore do not properly signal the animal's position.

Because these results suggested that fairly normal CA1 place cell activity is still possible when NMDA receptors are deleted from the CA1 region, we asked whether a more widespread blockade of NMDA receptors might cause a greater disruption of CA1 place cells. For example, does the formation of place cells in the CA1 region require normal NMDA-mediated LTP in other parts of the hippocampus or the neocortex? We therefore used global pharmacological blockade of NMDA receptors in all brain areas. A pharmacological blockade also offered the advantage of temporal control, allowing us to investigate the effects of NMDA receptor blockade on both the maintenance of a previously formed place cell map and the establishment of a new map in a novel environment.

Our results with acute, global interference with NMDA receptors confirm the main conclusion of McHugh et al. (19) that NMDA receptors must be available for place cells to be normal and extend that conclusion by showing more precisely the role played by NMDA receptors in the formation and long-term maintenance of a place cell map. However, our results also differ in one respect from those of McHugh et al. (19). Whereas they found that chronic knockout of NMDA receptors in CA1 resulted in CA1 place cells with somewhat enlarged and diffuse firing fields, we found that CPP had no effect on field size or quality. One possible explanation for this difference is that McHugh et al. (19) were able to eliminate NMDA receptor subunit 1 protein completely by genetic means, but the dose of CPP that we used may not have been sufficient to block NMDA receptors completely. However, the dose we used is twice as great as needed to impair spatial memory (20) and is sufficient to reversibly prevent primed-burst potentiation (Fig. 4) and hippocampal LTP (21). A second possibility is that in the study of McHugh et al. (19) the gene encoding the NMDA receptor subunit 1 protein was knocked out during a period in the development of the mapping system when NMDA receptor expression is still required for the formation of normal synaptic organization (22).

Consistent with previous studies on LTP and learning, we found that the hippocampal representation of an already familiar environment was unaffected by global NMDA receptor blockade [(9,10); see also (23)]. A surprising result of our study was that NMDA-dependent processes are also not required for creating new firing fields in a novel environment or for the short-term maintenance of the new fields, although they are required for long-term maintenance of the new fields. The maintained quality of firing fields in the familiar environment and the development of crisp new fields in a new environment suggest that the place cell system receives adequate sensory information despite blockade of NMDA receptors and therefore that the deficits caused by system-wide NMDA receptors blockade are not due to interference with sensory systems. Reliable location-specific firing in a new environment requires that cell activity becomes linked to a sensory configuration that exists only within the cell's firing field. How can this linking occur if NMDA receptors are blocked? Perhaps place cells are tuned to certain stimuli by genetic or developmental events before the rat enters the new environment. In this case, the resemblance of firing fields in the two environments might be expected to reflect the resemblance of the environments to each other. Remapping is, however, often complete even when a new environment closely resembles the old environment (6, 7). Moreover, it is hard to understand why preexisting tuning would allow the same pyramidal cells to have fields that are different during days 1 and 2 in the same environment, as happened with the CPP animals.

Our data, therefore, suggest the interesting possibility that, in addition to the NMDA-dependent plasticity essential for long-term stability of a new map, there exists a second, more labile and NMDA-independent form of plasticity that is sufficient to allow firing fields to form and to be maintained for 1.5 hours. The possibility that there are two forms of plasticity for different phases of spatial memory processes is consistent with topological mapping theories, which require firing fields to be established first by some unspecified mechanism, after which NMDA-dependent plasticity encodes the distance between the fields, creating a map that can be used to solve navigational problems (24). This more labile form of plasticity could be sufficient to subserve working memory in the radial arm maze, which persists for several hours during NMDA receptor blockade (25), and might also contribute to the ability of rats to learn other spatial tasks during blockade of NMDA receptors (26).

The most profound effect of NMDA receptor blockade was to disrupt the long-term (16 to 24 hour) stability of a newly formed firing field map: The first set of fields disappeared and was replaced by a second, newer set of fields the next day. McHugh et al. (19) did not investigate 24-hour stability. However, studies of several other types of mutant mice with deficits in both LTP and spatial learning have found that they also have as a common feature place fields with short- or long-term instability, although the fields in these mutant mice have other abnormal properties as well (27). Our results indicate that acute pharmacological blockade of NMDA receptor–dependent processes produces a selective deficit in long-term stabilization of new firing fields, with little effect on other firing field properties. Because some forms of both spatial learning and LTP are also NMDA- dependent (9,10), our results suggest that these three phenomena may be related: The same plasticity mechanisms that underlie the long-term maintenance of LTP may be required for long-term stabilization of new place field maps (either in hippocampus or in other brain areas), which in turn may be necessary for spatial memory.

  • * Present address: Center for Neurobiology and Behavior, College of Physicians and Surgeons, Columbia University, 722 West 168 Street, New York, NY 10032, USA.

  • To whom correspondence should be addressed. E-mail: bob{at}fasthp.hippo.hscbklyn.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article