Place Cells Can Flexibly Terminate and Develop Their Spatial Firing. A New Theory for Their Function

Transcription

1 PII S (99) Physiology & Behavior, Vol. 67, No. 1, pp , Elsevier Science Inc. Printed in the USA. All rights reserved /99/$ see front matter Place Cells Can Flexibly Terminate and Develop Their Spatial Firing. A New Theory for Their Function NANDOR LUDVIG 1 Department of Physiology and Pharmacology, State University of New York, Health Science Center at Brooklyn, Box 31, 450 Clarkson Avenue, Brooklyn, NY Received 30 July 1998; Accepted 11 February 1999 LUDVIG, N. Place cells can flexibly terminate and develop their spatial firing. A new theory for their function. PHYSIOL BE- HAV 67(1) 57 67, In this study, hippocampal place cells were recorded in a behavioral paradigm previously not employed in place-cell research. Rats were exposed to the same fixed environment for as long as 8 24 h without interruption, while the firing of CA1 and CA3 place cells was monitored continuously. The first finding was that all place cells that were detected at the beginning of the recording sessions ceased to produce location-specific firing in their original firing fields within 2 12 h. This was observed despite the fact that the animals kept visiting the original firing fields, the hippocampal EEG was virtually unchanged, and the discriminated action potentials of the cells could be clearly recorded. The second finding was that some complex-spike cells that produced no spatially selective firing pattern at the beginning of the recording sessions developed location-specific discharges within 3 12 h. Thus, place cells can flexibly terminate and develop their spatial firing, even in a fixed environment and during similar behaviors, if that environment is explored continuously for a prolonged period. To explain this phenomenon, a new place-cell theory is outlined. Accordingly, the high-frequency discharges of these neurons may serve to create, under multiple extrahippocampal control and within limited periods, stable engrams for specific spatial sites in the association cortex where the cognitive map probably resides. After the creation of a stable engram, or in the absence of favorable extrahippocampal inputs, place cells may suspend their location-specific firing in the original field, and initiate the processing of another spatial site Elsevier Science Inc. Rat Exploratory behavior Hippocampus Place cell Action potential Cognitive map IT was discovered in 1971 that during movement in space hippocampal principal cells can markedly increase their firing rate at specific locations (20). The properties of these neurons, named place cells, have been later elaborated by numerous investigators, opening a new chapter in hippocampus research (7,15,18,22). Along with the discoveries that surgical lesions of the hippocampus severely impair the ability to form new memories (26), and that intrahippocampal electrical stimulations can induce long-term potentiation (LTP) in the local synaptic transmission (2), the place-cell phenomenon seems to be the third major clue to the role of hippocampus in brain. Extensive studies notwithstanding, the biological laws that govern the function of place cells and transform their discharges into a neural product that represents space have remained elusive. The facts that these neurons are found primarily in the hippocampal formation, that most principal cells in this structure can function as place cells, and that different place cells fire in different locations ( firing fields ), undoubtedly give the impression that the network of these neurons is actually a cognitive map (15,18,21). But compelling evidence shows that the hippocampal formation is also involved in cognitive processes other than mapping space; its role in the formation of permanent declarative memory in the neocortex is especially well documented (27). Indeed, hippocampal principal cells can also increase their firing during classical conditioning (1), odor-discrimination tasks (6), and many other nonspatial paradigms. How can these diverse data be integrated into a theory where all pieces fit, and harmonize with the phenomenon of LTP? To respond this challenge, first some crucial aspects of place-cell firing should be reexamined. 1 To whom requests for reprints should be addressed. ludvin10@bmec.hscbklyn.edu 57

2 58 LUDVIG FIG. 1. The multifaceted spike discrimination employed in this study. The cluster analysis shows the X/Y plot of the first positive wave (valley, in mv) and the amplitude (peak, in mv) of all action potentials recorded with a microelectrode pair in a 10-min session. Arrowhead points to a distinct cluster; the overlaid waveforms of the action potentials in this cluster are indicated on the right. The firing rate distribution map demonstrates the spatial properties of the same action potentials in a cylindrical test chamber; note the wellcircumscribed firing field (arrowhead). The color code for the frequency of the action potentials is indicated below the map. The interspike interval histogram demonstrates that no action potential followed a preceding one within 4 ms, proving that the spatially selective discharges derived from a single place cell. The location-specific action potential volley shows the very spikes that were discharged inside the firing field. Note that the volley comprises slightly different waveforms. Negativity up, as in the other figures.

3 PLACE-CELL FIRING TERMINATION AND DEVELOPMENT 59 One of the unresolved problems of the location-specific firing of place cells is the stability of these discharges. It is clear that changing the physical structure of the environment, rearranging its significant cues, or introducing new local and distal landmarks can profoundly affect the firing of place cells (3,17,28). Alterations in the behavioral tasks performed in an environment can also be accompanied with alterations in place-cell firing (14). In all of these studies, however, either the environment or the behavior was changed, and the observed place-cell firing alterations were attributed to these experimental manipulations. When such manipulations have not been used, the firing of place cells have been demonstrated to be stable, even for months (29). Thus, it is generally accepted that during constant behavior in a fixed environment place cells increase their firing rate always at the same locations. This is why it is hypothesized that the hippocampal network of place cells is a cognitive map (21), or a cognitive graph (19), or part of a preconfigured network that generates an abstract internal representation of two-dimensional space (15). A close inspection of the experimental designs used in place-cell research reveals that there is a component that has been common, without exception, in all studies. Namely, all studies have utilized brief, min, recording sessions separated by long intersession intervals. For these intersession intervals, the animals were removed from the test area. Thus, they have never been allowed to examine the experimental Cell No. TABLE 1 SUMMARY OF THE FIRING PATTERN CHANGES IN THE RECORDED PLACE CELLS Localization Spatial Firing Change Occurrence of Change I. Cells that displayed spatial firing at the beginning of the session 1 CA3 Termination, then 6th hour background firing (Fig. 2) 2 CA1 Termination, then 9th hour background firing (Fig. 3) 3 CA3 Termination, then 7th hour background firing 4 CA1 Termination, then 2nd hour background firing 5 CA3 Termination, then 3rd hour background firing 6 CA3 Termination, then 12th hour background firing* 7 CA3 Termination, then new 2nd hour spatial firing (Fig. 4) 8 CA1 Termination, then no action potentials 5th hour II. Cells that did not display spatial firing at the beginning of the session 9 CA1 Development of spatial 10th hour firing 10 CA1 Development of spatial 3rd hour firing 11 CA3 Development of spatial 6th hour firing (Fig. 5) 12 CA3 Development of spatial firing 12th hour *This cell retained spatial firing in a small secondary firing field. This refers to the termination of spatial firing in the original firing field. The location-specific discharges developed at the new site in the 3 rd hour, and then were stable until the end of the 24-h session. environment for long periods undisturbed and without interruption, at their own pace, with food, water, and time for rest and sleep provided. But in life, it often occurs that an animal or human explores a fairly stable environment (e.g., a home territory, a mating place, a hunting area, etc.) for long periods. Furthermore, it is possible that the formation of the cognitive map runs on a different course during such circumstances than during brief, interrupted environmental encounters. Yet, the stability of hippocampal place cells during long exposure to a fixed environment has not been investigated. The goal of the present study was, therefore, to answer the basic question: What happens to the firing of place cells during uninterrupted, prolonged exposure to the same environment? The experiments revealed that these neurons can flexibly terminate and develop their spatial firing. Because this does not fit into the classic notion that hippocampal place cells form a cognitive map, a new place-cell theory will be proposed. Some of the data and ideas of this article were presented in an abstract form (13). MATERIALS AND METHODS Animals and Behavioral Training In this study, nine male hooded Long Evans rats were used. All of them were young animals with their body weight ranging from 300 to 400 g. The rats were kept on a 12/12 hour light/dark cycle with lights on at 0700 h. The rats were trained to chase food pellets scattered on the floor of a test chamber (16). Electrodes and Surgical Procedures Driveable microelectrode assemblies were used. They were similar to those employed in our previous place-cell studies (10,12) except that this time microdialysis probe was not attached to the electrodes. Each assembly contained an array of 8 microwires (Nichrome; 25 m diameter). Importantly, the device was equipped with a stainless steel hook, through which it could be firmly secured to the recording cable. This assured that the cable was never disconnected during the prolonged recording sessions. Moreover, custom-made driving screws were used, of which neck was embedded in a nylon block. These blocks could hold the driving screws very tightly, providing very stable electrode positions in the brain. The microelectrodes were implanted in the brain chronically. The rat was anesthetized with 50 mg/kg pentobarbital, i.p., and placed in a stereotaxic apparatus. The skull was exposed, and four anchoring screws were placed in the frontal and occipital bones. Then a 2-mm diameter hole was drilled above the hippocampus. The coordinates of the center of this hole were: 3.3 mm posterior to the bregma and 2.5 mm lateral to the midline. The microelectrode array was lowered stereotaxically into the sensorimotor cortex, 1.7 mm below surface. The grounding wire was connected with Silver Print conductive paint to one of the anchoring screws above the cerebellum, and the assembly was anchored to the skull with dental cement. The Experimental Paradigm The recording sessions were conducted 4 8 days after the surgery. The animal was placed into a small box in the center of a sound-attenuating, ventilated Faraday cage (240 cm height 135 cm length 135 cm width) and was connected to

4 60 LUDVIG FIG. 2. The phenomenon of place-cell firing termination. The data were collected in a rat, which was behaving freely in a cylindrical test chamber for 8 h without interruption. The environment was fixed, and the cylinder was familiar to the animal. Each firing rate distribution map shows the average firing rate of cell 1 in the areas of the pixels, as the rat moved around during a 15-min data collection period. The color code for the firing rates is indicated below the maps. In the first recording hour, the cell increased its electrical activity in a well-circumscribed firing field (arrowhead), thus functioned as a place cell. However, in the sixth hour this firing pattern disappeared. The overlaid action potentials of this cell, as well as those of a neighboring neuron (cell 2) recorded with the same electrodes, prove stable electrophysiological recording throughout the experiment. Calibrations as indicated. The staying time distribution maps demonstrate that the rat displayed similar movement patterns in the first and the sixth hours. the recording cable. The microelectrodes were advanced toward the hippocampus by rotating the driving screws. When clear pyramidal/complex-spike cells (5,8,25) were detected, the rat was placed intermittently for 10-min periods into the cylinder (70 cm diameter; 50 cm height) to test whether place cells are present in the recordings. If yes, data collection started. The place cells were monitored either in the cylinder for 8 h or in a rectangular chamber (50 cm height 90 cm length

5 PLACE-CELL FIRING TERMINATION AND DEVELOPMENT 61 Cell No. TABLE 2 PEAK FIRING RATES IN THE AREA OF THE ORIGINAL FIRING FIELD Firing Rate (Hz) in the 1st Hour of the Recording Session 90 cm width) for as long as 8 24 h. Because these latter experiments included overnight monitoring, the rectangular chamber was equipped with a water bottle. In two experiments, a pyramid-shaped structure was placed in the center of this chamber to form a rectangular track. Following one of these rectangular track sessions, the rat was placed back to the cylinder and the firing of the monitored place cell was tested again for 10 min in this environment. Food pellets were provided to the rat with an automatic pellet dispenser hourly, for min periods. The illumination of the Faraday cage was switched off between 2100 and 0700 h; otherwise the interior of the test area was unchanged. Five elements of this paradigm should be emphasized. First, each animal was exposed to an experimental environment continuously for a longer period than what has been used previously in place-cell research. During these prolonged periods, the rats were allowed to freely explore the environment, foraging for food pellets, rest, sleep, and in the rectangular chamber: drink. Second, by using three different test areas (cylinder, rectangular chamber, and rectangular track), it was possible to determine whether the observed place-cell firing changes are characteristic only to a specific environment. Third, the environment was fixed in each experiment. Fourth, by providing food pellets only periodically, the behavior of the rats could be controlled. This allowed the collection of place-cell data during very similar movements. Fifth, while the rectangular chamber and the track were novel environments to the rats, the animals became familiar with the cylinder during the initial testing of place cells. This provided an opportunity to monitor the firing of place cells in both novel and familiar environments. The described long-term recording sessions were performed on six rats. The remaining three rats were subjected only to brief, 10-min recording sessions in the cylinder. One of these rats was retested in the cylinder after a 20-h intersession interval. During this interval, the animal rested in the small box. Electrophysiological Data Acquisition Firing Rate (Hz) 3 12 h Later* Mean SEM p 0.01 (t = 4.283; df = 5) *For exact times see the Occurrence of change column in Table 1. The recording of place cells was conducted in a similar manner as in our previous studies (10,12). One of the eight microelectrodes was used as a reference (noninverting) electrode, and the rest of the microelectrodes served as active (inverting) electrodes. The action potential-related extracellular signals were fed into impedance-lowering TLC2274 operational amplifiers (Texas Instruments, Dallas, TX). These devices effectively eliminated the movement artifacts from the recordings. The hippocampal electroencephalogram (EEG) was also recorded, using one of the microelectrodes as the active electrode and the grounding screw as the reference electrode. The electrical signals were fed, via a commutator, into differential AC amplifiers. The EEG waves were amplified 1000 times, filtered between 0.1 Hz and 50 Hz, displayed on an oscilloscope, and occasionally printed. The cellular signals were amplified 10,000 times, and filtered between 300 Hz and 10,000 Hz. Then they were digitized at a sampling rate of 40,000 Hz, and collected on the hard disk of a computer with the use of the Discovery data acquisition software (Data- Wave Technologies, Longmont, CO). The voltage threshold for the acquisition of the incoming spikes was set to a level that was approximately two times higher than the voltage of the V noise but sufficiently lower than the voltage of the V cellular signals. During the prolonged recording sessions, min data files were generated continuously in two rats, while in four rats 15-min data files were collected in every 1 4 h. Behavioral Data Acquisition The head position of the animal was continuously recorded with a light-emitting diode (LED) tracker system. This system consisted of 1) a pair of red LEDs mounted on the operational amplifiers, 2) a NC-16TC CCD color camera, and 3) the LED tracker (Ebtronics, Elmont, NY). The tracker divided the video field of the camera to 256 X and 242 Y coordinates, detected the pair of LEDs as the brightest spot in the field, and transferred the X and Y coordinates of the LEDs to the computer at a 60-Hz speed. These data were also acquired with the Discovery program. Thus, the recording apparatus captured not only the extracellular signals, but also their occurrence in space as the rat moved around in the test chamber. Histology and Data Analysis After the end of the experiments, the animals were euthanized with 120 mg/kg pentobarbital, i.p. Their brain was removed, and immersed in liquid N 2. The frozen brains were sectioned with a cryostat at 10 C to obtain 50 m-thick coronal sections. The sections were stained with cresyl violet, and viewed with a light microscope to determine whether the electrode tips were localized in the hippocampus. In two experiments the histological procedures were preceded by passing a current (15 20 A for 30 s) between two of the microelectrodes to make small electrolytic lesions. This article is based on the analysis of 15 place cells recorded from either the CA1 or the CA3 area. Twelve cells were recorded in the prolonged experimental sessions; three cells were recorded in the brief recording sessions. The offline discrimination of the collected action potentials was achieved with a multifaceted five-step spike analysis (Fig. 1). The detected waveforms were first subjected to cluster-cutting with the Spike Sort program of DataWave Technologies. This program combines various parameters of the action potential waveforms and displays them in a X/Y point plot format. Then the action potentials belonging to a cluster were displayed and visually examined. Next, the spatial properties of the firing of the discriminated neuron were analyzed with the Mapmaker program (ESCO, Mt. Kisco, NY). This program used both the discriminated spikes and the head-position data, and generated firing-rate distribution maps. Briefly, the Mapmaker first divided the entire test area available for

6 62 LUDVIG FIG. 3. The relationship between place-cell firing termination and the continuous, prolonged nature of the environmental exposure. The data were collected in a freely behaving rat, during its brief 10-min stays in a familiar cylinder before and after the continuous, 24-h exposure to a novel rectangular track. Both environments were fixed. Each map shows the average firing rate of a place cell in the areas of the pixels, as the rat moved around. The data collection periods were 10 min in the cylinder and 50 min in the rectangular track. The color code for the firing rates are indicated below the maps. Note that in the rectangular track (northern wall) the location-specific discharges developed within the first hour, were unchanged in the fifth hour, but started to occur diffusely in the seventh hour. By the 19th hour, the cell ceased to increase its electrical activity in the original firing field, and produced only a background firing of Hz. In contrast, the cell retained its spatial firing in the cylinder (southern wall).

7 PLACE-CELL FIRING TERMINATION AND DEVELOPMENT 63 FIG. 4. Place-cell firing termination followed by the development of spatial discharges at a new location. The data were collected in a rat, which was behaving freely in a rectangular chamber for 24 h without interruption. The environment was fixed, but the chamber was novel to the rat. Each firing rate distribution map shows the average firing rate of a place cell in the areas of the pixels, as the rat moved around in the chamber during a 15-min data collection period. The color code for the firing rates are indicated below the maps. Note the rapid development of spatial firing in the first hour in the center of the test chamber, and that 3 h later this firing pattern completely disappeared. However, the neuron developed location-specific discharges at a new location (top right corner). This firing pattern change was gradual. The overlaid action potential waveforms obtained in the first and the third hour indicate that these discharges derived from the same cell. Also note the similarities of the staying time distribution maps. the rats into 28 by 28 pixels, and calculated the total time (seconds) the experimental animal spent in each pixel during a given data collection period. Then the program calculated the total number of action potentials that were discharged in each of these pixels, as the rat moved around. The average firing rate for each pixel was calculated by dividing the number of action potentials with the staying time. These average firing rates were ranked according to their values and assorted into

8 64 LUDVIG FIG. 5. Development of place-cell firing in a nonspatial complex-spike cell (cell 2 in Fig. 2) during the sixth hour of the environmental exposure. Each firing rate distribution map shows the average firing rate of the neuron in the areas of the pixels, as the rat moved around in a cylinder during a 15-min data-collection period. See description of the experiment, action potentials waveforms, and corresponding staying time distribution maps in Fig. 2. Note that in the first recording hour the cell was virtually silent, but 5 h later it developed clear location-specific discharges, just off the center of the test area. Cell No. TABLE 3 PEAK FIRING RATES IN THE AREA OF THE DEVELOPED FIRING FIELD Firing Rate (Hz) in the 1st Hour of the Recording Session Firing rate (Hz) 3 12 h Later* Mean SEM p 0.05 (t = 3.684; df = 4) *For exact times see the Occurrence of change column in Table 1. five groups. The mean firing rate for each group was calculated, and indicated with a color; yellow was added to indicate zero firing rate. The maps were generated by filling each pixel with the appropriate color, with the empty areas indicating locations the rat did not visit. To assure that the firing rate distribution maps truly represented the discharges of a single neuron, the action potentials were subjected to interspike interval histogram analysis. Because the minimal refractory period of hippocampal complex-spike/place cells is approximately 2.5 ms, the analyzed action potentials were considered to derive from a single neuron only if the minimal interspike interval was not below this value. The multifaceted data analysis was completed by extracting the very spikes that were discharged in the firing fields. These location-specific action potential volleys were displayed and compared. In addition, the Mapmaker program also generated a staying time distribution map for each data file. These maps solely used the total time the rat spent in each pixel. The time values were assorted into six groups, and the mean staying time for each group was calculated. The mean staying times were indicated with a color. Again, the maps were generated by filling each pixel with the appropriate color, with the empty areas indicating locations the rat did not visit. These maps provided important information on the movement pattern of the animal. Finally, the printed EEG waves were visually examined. Statistical analysis was performed on the data collected in the 8 24-h recording sessions. For this analysis, the firing rate distribution maps were used. First, the firing field of each place cell was manually circumscribed in the maps. Then the peak firing rates within these areas were identified. Peak firing rate was defined as the mean of the highest firing rate range, indicated with the black pixels. For each cell, two peak firing rates were selected: one that had been displayed in the first recording hour, and another one which had been displayed 2 12 h later. These values were analyzed with paired t-test. RESULTS The used multifaceted spike analysis provided appropriate waveform discrimination (Fig. 1). With this spike discrimination method, 12 place cells were isolated in the 8 24-h recording sessions. The localizations of these neurons are indicated in Table 1. During the prolonged recording periods, the rats were exploring the environment, foraging for food, resting, sleeping, drinking, and grooming. When foraging for food, the animals displayed fairly similar movement patterns. This allowed the collection of the place-cell data during constant behaviors. The prolonged, uninterrupted environmental exposures induced profound changes in the electrical activity of the place cells. These changes and their occurrence time are summarized in Table 1. A total of eight place cells were detected at the beginning of the 8 24-h data collection periods. The firing of one cell rapidly diminished in the fifth hour and then disappeared an hour later. This indicated that the electrodes either damaged the neuron or drifted far away from it. However, such technical problems did not occur in the case of the other seven place cells. It was found that all of these seven neurons terminated their location-specific discharges in the original firing field, gradually, within 2 12 h. In the case of six cells, the initially robust spatial firing blended into a diffuse background firing. This phenomenon is demonstrated in Fig. 2. The termination of the location-specific discharges was reflected in a statistically significant reduction of the peak firing

9 PLACE-CELL FIRING TERMINATION AND DEVELOPMENT 65 rates in the original firing field (Table 2). Figure 3 illustrates the time course of place-cell firing termination, as well as the relationship between this phenomenon and the uninterrupted, prolonged nature of the environmental exposure. In this experiment, the monitored place cell lost its ability to generate location-specific discharges in the long-exposure environment (rectangular track), but continued to produce spatial firing in the brief-exposure environment (cylinder). Following place-cell firing termination, a slow development of location-specific discharges in a new site was also observed (Fig. 4). Interestingly, a slow development of location-specific discharges were also detected in four complex-spike cells, which did not display spatial firing at the beginning of the experiments. As the prolonged recording sessions proceeded, these neurons were transformed to place cells. This phenomenon is shown in Fig. 5. The firing rate increases in the newly developed firing fields were statistically significant (Table 3). The staying time distribution maps in Figs. 2 and 4 indicate that the movement patterns of the animals were very similar before and after the place-cell firing changes. The overlaid action potential waveforms (Figs. 2 and 4) demonstrate that the spikes of the discriminated cells were safely recorded throughout the recording sessions. The hippocampal EEG recordings showed that place-cell firing termination could occur in the presence of virtually unchanged theta rhythm (Fig. 6). The three place cells that were recorded during the brief exposures to the cylinder fired in the same way as the above-described cells did at the beginning of the prolonged recording sessions. The place cell that was retested in the cylinder after a 20-h intersession interval displayed stable spatial firing. DISCUSSION In agreement with previous studies by other investigators (3,7,14 17,20,23,28,30), this report confirms that place cells can be readily detected in the rat hippocampus, and that these neurons can rapidly develop long-lasting spatial discharges in a new environment. However, the novel paradigm employed in the present study also revealed two phenomena that have not been documented before. Namely, place cells can flexibly terminate and develop their spatial firing in a fixed environment if that environment is explored continuously for a prolonged period. As Figs. 2, 4, and 6 demonstrate, these firing pattern changes are not due to behavioral or EEG alterations. Necessarily, these experiments utilized extracellular recording in freely moving animals. However, this technique is bedeviled by several complicating factors. The penetration of the electrodes can cause extensive cellular damage and abnormal local blood flow. Vigorous head movements and cerebrovascular pulsation can push the electrodes either away from or closer to the recorded neurons, as well as into different vertical positions along the dendritic axis. Inflammatory processes and reactive gliosis can affect the conductance of the extracellular medium, and activation of silent cells can interfere with the potential fields of the monitored neuron. These factors are impossible to control. Therefore, it cannot be excluded that to some extent the data of this study were confounded by unwanted and unrecognized recording artifacts. Nevertheless, the clearly stable recordings shown in Figs. 2 5 suggest that it is very likely that the experiments did uncover neurobiologically significant cellular processes. How can the described place-cell firing termination be explained? The classic place-cell theories (15,18,21) postulate that the network of these neurons forms a cognitive map. It is possible that a specific location in a given environment is originally coded by a large number of place cells, but as the exploration proceeds, some of these neurons terminate their spatial firing. Furthermore, it is possible that place-cell firing termination is merely a transient event. But this phenomenon also supports the idea that hippocampal place cells may not form a cognitive map; rather, they promote within limited periods its formation in extrahippocampal areas (11). This theory is summarized in Fig. 7. Essentially, the following is proposed: during the visit of a firing field, all stimuli from the body and the FIG. 6. Evidence that the phenomenon of place cell firing termination is not the consequence of diminished hippocampal theta rhythm. Hippocampal EEG recordings are shown, as they were recorded between a microelectrode and the ground, during the course of the experiment shown in Fig. 3. The upper trace was recorded in the first hour in the rectangular track, during one of the moments when the rat crossed the firing field. The lower trace was recorded in the 20th hour in the rectangular track, during one of the moments when the rat crossed the original firing field of the same cell, but after the termination of its location-specific discharges. The EEG recordings demonstrate that in both cases the hippocampus generated very similar, 5 7 Hz theta waves. Calibrations as indicated.

10 66 LUDVIG space are first transmitted to the primary sensory cortical areas. The primary sensory cortical neurons send action potentials to neurons in the association cortex. The association cortical neurons, stimulated within the same time window, form an unstable cell assembly, which is actually a nascent memory unit or engram. This engram will vanish unless its cellular components manage to send action potentials, via the entorhinal cortex, to the hippocampal place cells. The place cells, which themselves are under multiple extrahippocampal control, react to these inputs with the generation of LTP. As a consequence, the entorhinal inputs are potentiated, enabling the place cells to send back high-frequency discharges to the same cell assembly. These discharges strengthen the synaptic connections within the assembly and initiate the consolidation of the engram. The operation of this place cell-association cortical cell loop lasts until the engram creation is in process. When it is completed, the place cells become temporarily inactive. Accordingly, place cells may terminate their spatial discharges in the original firing field either when they receive unfavorable control inputs from extrahippocampal areas or FIG. 7. A theory for the role of hippocampal place cells. (A) The hippocampal place cell, as a neuronal device to strengthen the synaptic connections within an association cortical cell assembly, creating a permanent engram for a spatial site. Note, that this proposal incorporates the role of cerebral cortical cell assemblies in memory formation (9), consistent with the data that suggest that the hippocampus is principally involved in the formation of new memories in the neocortex (27), agrees with the idea that pyramidal cell bursts in the hippocampus can promote memory formation (4,24), and handles the mechanism of LTP as a prerequisite for normal place cell function. (B) The place cell-created spatial engrams, as the building blocks of the association cortical cognitive map. Note that these engrams are chained together in the temporal order of their creation, thus allowing to code time, and are organized into a larger unit by a reference engram. This larger unit is the representation of an environment. Synaptic connections between such larger units make up the cognitive map. Distance between two spatial sites can be coded by the number of engrams that separates the engrams for these spatial sites; direction can be coded in the relation of the individual engrams to the reference engram. The system operates in the same way when movement in space does not occur, but in this case each engram codes a sequence of sensory stimuli that are organized by the reference engram into the representation of an episode. This is consistent with the involvement of hippocampal pyramidal cells in nonspatial cognitive processes (1,6,7). Indeed, the outlined theory views the cognitive map as only one manifestation of a single memory system, where space and time are both coded by the same engram-creating mechanism.

11 PLACE-CELL FIRING TERMINATION AND DEVELOPMENT 67 when they complete the creation of a permanent associational cortical engram for that location. This does not contradict the finding that by the employment of brief, interrupted recording sessions stable place-cell firing can be recorded even for months (29). In this paradigm the exploration of the environment is discontinued, preventing the completion of spatial engrams within a single experimental session. As a consequence, the place cells are repeatedly engaged in the engram-creating process, and thus appear in the recordings as neurons that produce stable location-specific firing. Indeed, the cell in Fig. 3 displayed stable location-specific firing in the cylinder, where the rat stayed only for 10-min periods. The development of place-cell firing in this study (Figs. 4 and 5) probably reflected the transition of inactive place cells to neurons that regained their ability to process a spatial engram for a new environmental location. In fact, the gradual termination and development of place-cell firing seem to be complementary mechanisms, both contributing to the continuous updating, refining, and expansion of the association cortical cognitive map. Such dynamic firing pattern changes would explain why the frequency of place-cell firing varies so much from cell to cell and session to session (14,23,28). Further studies are required to determine whether all place cells are capable of terminating their spatial discharges in the original firing field or just a subclass of place cells. Also, it has yet to be shown whether the described place-cell firing termination is a transient or a permanent phenomenon. Nevertheless, the present experiments suggest that the electrical activity of hippocampal place cells is even more flexible than previously thought. This also has a practical aspect: pharmacological studies involving place cells must recognize and separate the spontaneous and drug-induced place-cell firing changes. ACKNOWLEDGEMENTS I wish to express my gratitude to software designer Lorant Kovacs, President of ESCO, for his essential contribution to the development of the Mapmaker program, and to Dr. Hai Michael Tang for his superb technical assistance. I am indebted to Dr. Steven E. Fox for introducing me into the art of extracellular recording, and to Dr. John L. Kubie for his help to set up the DataWave system. The encouragements of Dr. James B. Ranck, Jr., throughout the course of this study, were invaluable. This work was supported by NIH Grants AA10814 and MH56800, and by a Research Investment Initiative Grant from SUNY HSCB. 1. Berger, T. W.; Thompson, R. F.: Identification of pyramidal cells as the critical elements in hippocampal neuronal plasticity during learning. Proc. Natl. Acad. Sci. USA 75: ; Bliss, T. V. P.; Lomo, T.: Long-lasting potentiation of synaptic transmission in the dentate area of the anaesthetized rabbit following stimulation of the perforant path. J. Physiol. 232: ; Breese, C. R.; Hampson, R. E.; Deadwyler, S. A.: Hippocampal place cells: Stereotypy and plasticity. J. Neurosci. 9: ; Buzsaki, G.: Two-stage model of memory trace formation: a role for noisy brain states. Neuroscience 31: ; Buzsaki, G.; Penttonen, M.; Nadasdy, Z.; Bragin, A.: Pattern and inhibition-dependent invasion of pyramidal cell dendrites by fast spikes in the hippocampus in vivo. Proc. Natl. Acad. Sci. USA 93: ; Echenbaum, H.; Kuperstein, M.; Fagan, A.; Nagode, J.: Cue-sampling and goal-approach correlates of hippocampal unit activity in rats performing an odor-discrimination task. J. Neurosci. 7: ; Eichenbaum, H.: Is the rodent hippocampus just for place? Curr. Opin. Neurobiol. 6: ; Fox, S. E.; Ranck, J. B., Jr.: Electrophysiological characteristics of hippocampal complex-spike cells and theta cells. Exp. Brain Res. 41: ; Hebb, D. O.: The organization of behavior. New York: Wiley & Sons; Ludvig, N.; Chao, K.; Altura, B. T.; Altura, B. M.; Fox, S. E.: Manipulation of pyramidal cell firing in the hippocampus of freely behaving rats by local application of K via microdialysis. Hippocampus 6:97 108; Ludvig, N.: Microdialysis-coupled place cell detection on the hippocampus: A new strategy for the search for cognition enhancer drugs. Prog. Neuropsychopharmacol. Biol. Psychiatry 21: ; Ludvig, N.; Fox, S. E.; Kubie, J. L.; Altura, B. M.; Altura, B. T.: Application of the combined single-cell recording/intracerebral microdialysis method to alcohol research in freely behaving animals. Alcohol. Clin. Exp. Res. 22:41 50; Ludvig, N.: Hippocampal place cells as engram-creating neuronal devices. Soc. Neurosci. Abstr. 24:1416; Markus, E. J.; Qin, Y.; Leonard, B.; Skaggs, W. E.; McNaughton, B. L.; Barnes, C. A.: Interactions between location and task affect the spatial and directional firing of hippocampal neurons. J. Neurosci. 15: ; McNaughton, B. L.; Barnes, C. A.; Gerrard, J. L.; Gothard, K.; REFERENCES Jung, M. W.; Knierim, J. J.; Kudrimoti, H.; Qin, Y.; Skaggs, W. E.; Suster, M.; Weaver, K. L.: Deciphering the hippocampal polyglot: The hippocampus as a path integration system. J. Exp. Biol. 199: ; Muller, R. U.; Kubie, J. L.; Ranck, J. B., Jr.: Spatial firing patterns of hippocampal complex-spike cells in a fixed environment. J. Neurosci. 7: ; Muller, R. U.; Kubie, J. L.: The effects of changes in the environment on the spatial firing of hippocampal complex-spike cells. J. Neurosci. 7: ; Muller, R.: A quarter of a century of place cells. Neuron 17: ; Muller, R. U.; Stead, M.; Pach, J.: The hippocampus as a cognitive graph. J. Gen. Physiol. 107: ; O Keefe, J.; Dostrovsky, J.: The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Res. 34: ; O Keefe, J.; Nadel, L.: The hippocampus as a cognitive map. Oxford: Clarendon Press; O Keefe, J.: A review of the hippocampal place cells. Prog. Neurobiol. 13: ; O Keefe, J.; Burgess, N.: Geometric determinants of the place fields of hippocampal neurons. Nature 381: ; Otto, T.; Eichenbaum, H.; Wiener, S. I.; Wible, C. G.: Learningrelated patterns of CA1 spike trains parallel stimulation parameters optimal for inducing hippocampal long-term potentiation. Hippocampus 1: ; Ranck, J. B., Jr.: Studies on single neurons in dorsal hippocampal formation and septum in unrestrained rats. Exp. Neurol. 41: ; Scoville, W. B.; Milner, B.: The loss of recent memory after bilateral hippocampal lesions. J. Neurol. Neurosurg. Psychiatry 20:11 21; Squire, L. R.; Zola Morgan, S.: The medial temporal lobe memory system. Science 253: ; Tanila, H.; Sipila, P.; Shapiro, M.; Eichenbaum, H.: Brain aging: Impaired coding of novel environmental cues. J. Neurosci. 17: ; Thompson, L. T.; Best, P. J.: Long-term stability of the place-field activity of single units recorded from the dorsal hippocampus of freely behaving rats. Brain Res. 509: ; Wilson, M. A.; McNaughton, B. L.: Dynamics of the hippocampal ensemble code for space. Science 261: ; 1993.