Instability in the Place Field Location of Hippocampal Place Cells after Lesions Centered on the Perirhinal Cortex

Abstract

The perirhinal cortex appears to play a key role in memory, and the neighboring hippocampus is critically involved in spatial processing. The possibility exists, therefore, that perirhinal–hippocampal interactions are important for spatial memory processes. The purpose of the present study was to investigate the contribution of the perirhinal cortex to the location-specific firing (“place field”) of hippocampal complex-spike (“place”) cells. The firing characteristics of dorsal CA1 place cells were examined in rats with bilateral ibotenic acid lesions centered on the perirhinal cortex (n = 4) or control surgeries (n= 5) as they foraged in a rectangular environment. The activity of individual place cells was also monitored after a delay period of either 2 min, or 1 or 24 hr, during which time the animal was removed from the environment.

Although the perirhinal cortex lesion did not affect the place field size or place cell firing characteristics during a recording session, it was determined that the location of the place field shifted position across the delay period in 36% (10 of 28) of the cells recorded from lesioned animals. In contrast, none of the place cells (0 of 29) recorded from control animals were unstable by this measure.

These data indicate that although the initial formation of place fields in the hippocampus is not dependent on perirhinal cortex, the maintenance of this stability over time is disrupted by perirhinal lesions. This instability may represent an erroneous “re-mapping” of the environment and suggests a role for the perirhinal cortex in spatial memory processing.

MATERIALS AND METHODS

Surgical procedures. Nine male Sprague Dawley rats weighing between 300 and 400 gm were anesthetized with sodium pentobarbitol (60 mg/kg, i.p.) and placed in a Kopf stereotaxic apparatus such that bregma and lambda were in the same horizontal plane. Body temperature was maintained at 37°C, and a midline incision was made to expose the skull.

Rats were divided into three groups and received either bilateral ibotenic acid (IBO) lesions of the perirhinal cortex (lesion;n = 4), no lesions (n = 3), or sham lesions (n = 2). In the treatment and sham animals, holes were drilled in the skull at 4.5, 5.5, and 6.5 mm posterior to bregma and 5.5 mm lateral to the midline. A guide tube (23 gauge needle) containing an obturator (30 gauge needle with sealed end) that extended (1.1 mm beyond the end of the guide tube, was then angled 10° laterally and located in the perirhinal cortex at a depth of ≈4.5–5 mm from the dural surface. In the case of the treatment animals, the obturator was withdrawn from the guide tube, the microinfusion cannula was inserted, and 0.3 μl of IBO (dissolved in a PBS, pH 7.4, at a concentration of 10 μg/μl) infused into each site over 4–5 min using a 5 μl syringe (5BR-5-RA8, SGE) connected to an automatic syringe pump (Bee Syringe Pump, MF-9090, BAS). The microinfusion cannula consisted of a 30 gauge dental needle (containing enough IBO for the whole surgery) that extended ≈1 mm beyond the end of the guide tube, attached to a length of plastic tubing filled with distilled water, and separated from the IBO by a 0.3 μl air bubble. The microinfusion cannula was ≈0.1 mm shorter than the obturator to ensure that it did not become blocked by tissue situated at the end of the guide tube. The cannula was left in the brain for 5 min after each infusion to allow for diffusion of the drug and then withdrawn from the guide tube and wiped down with a cotton swab. In the case of the sham lesions, the guide tube containing the obturator was inserted at each infusion site without any infusions being made.

A unit recording electrode consisting of a bundle of 6–7 (25 μm) formvar-coated nichrome wires mounted in a moveable “Scribe” microdrive (Bilkey and Muir, 1999) was then unilaterally (n = 7) or bilaterally (n = 2) implanted in the hippocampus at coordinates based on Paxinos and Watson (1998) (3.8 mm posterior to bregma, 2.5 mm lateral, ≈1.8 mm ventral). In unilateral surgeries, the implanted hemisphere was counterbalanced across animals. A stainless steel screw attached to the skull functioned as the ground. The area between the microdrive and the skull was sealed with Vaseline, and the microdrive was cemented in place with dental acrylic. A protective plastic ring was then cemented in place around the base of the microdrive. Stitches were placed anterior and posterior to the assembly, and penicillin was injected into the surrounding tissue. The rat was then placed in a heated recovery box and left to awaken from the anesthetic.

Apparatus and behavioral procedures. The experimental chamber consisted of a black rectangular chamber (120 × 60 × 60 cm) with a metal floor and an open top, within which the rat was allowed to move freely. Surrounding the chamber was a sheet of black polythene that loosely followed the lines of the chamber walls almost to the ceiling. Two back-lit cues were positioned in clearly visible locations on the inner walls of the chamber; a large crescent shape was positioned approximately one-quarter of the way along the long wall at a height of 35 cm, and a smaller star shape was positioned centrally on the end wall nearest the crescent at a height of 45 cm. After at least 10 d recovery after surgery, rats were habituated to the experimental chamber for 30 min/d and put on a schedule of food deprivation until they reached ∼85% of their free-feeding weight (which was then maintained for the duration of the experiment). During the period of habituation, animals learned to forage throughout the chamber to recover chocolate-flavored pellets placed randomly on the floor of the apparatus.

Animals were carried directly to the experimental room in an open top box. The room light was extinguished immediately before placing the animal in the chamber (always from the same location in the room, facing the star cue). When at least one unit with a satisfactory signal-to-noise ratio was isolated, a baseline recording session was begun. Each recording session lasted until 10–30 min of data had been recorded, during which time chocolate-flavored pellets were distributed throughout the experimental chamber to keep the rat moving. The unit was then recorded again after 2 min, 1 hr, and 24 hr delays with the order of these delays counterbalanced for different units. A unit recorded over all delays, therefore, was recorded over a 2 d period. The animal was returned to its home cage during the delay period for all delay durations, and the box was wiped clean. Only units that demonstrated (1) a low (<4 Hz) mean firing rate (FR) (averaged over all sessions), (2) a maximum firing rate (averaged over all sessions) of <30 Hz, and (3) a clear place field (PF) based on observation of the unit firing and location and behavior of the animal during a session, were categorized as place cells. In addition, almost all of these cells exhibited a peak-to-peak spike width of >400 μsec (Fox and Ranck, 1981). Furthermore, because it was important to ensure as much as possible that the same unit was being recorded in repeated sessions, only units that exhibited stable spike amplitudes and consistent waveforms within and between sessions were included in the present analysis. In most instances, the waveform characteristics of a given unit were consistent enough within and between sessions that the same cluster boundaries could be used in Datawave Discovery to uniquely identify the unit across all sessions. On some rare occasions in which the identity of the unit was not in question, some small adjustments were made to the cluster box of a unit to maintain accurate isolation. At the end of this procedure, or if no unit was present, the electrode was lowered ≈20–40 μm, and the animal was returned to its home cage until the next day. Once the electrode had been lowered and a “new” unit isolated, any similarities in waveform characteristics or place field properties between the new unit and an immediately preceding unit would result in the unit not being considered new and the electrode being lowered again.

After presentation of all the delay sessions, some units were recorded during additional cue rotation sessions. These rotation sessions were normally conducted on the day after the last delay session, although sometimes this involved two additional days of recording. For these sessions, animals were treated as for the delay sessions except that they were usually disoriented before entry to the recording chamber by covering and slowly rotating the transport box for 1 min as the experimenter randomly traversed the laboratory. Cues were manipulated by 180° rotations of (1) the visual cues only, (2) the recording chamber only (i.e., visual cues in the normal location relative to the rat's entry point), or (3) both the recording chamber and visual cues.

Electrophysiological procedures. All signals were amplified and filtered (300 Hz-3 kHz) by preamplifiers (Grass P511K) after being impedance-matched through a field effect transistor headstage. During unit recording, a separate channel of the unit electrode with minimal activity was used as the recording indifferent. An overhead video camera monitored the animal's movement by tracking the position of two different-colored LEDs mounted ∼7 cm apart on the headstage. This allowed the simultaneous acquisition of both the location of the animal and its head direction via a computer and custom tracking software (D. Bilkey) that generated DC outputs representing the x andy coordinates of the animal's position. All data were then digitized and recorded to tape for subsequent off-line analysis.

Histological procedures. At the conclusion of the experiment, the final electrode locations were marked by electrolytic lesions. Animals were then perfused with an infusion of saline (0.9%), followed by a 10% formalin solution. The brain was removed and placed in a 10% formalin solution for at least 1 d before being transferred to a 10% formalin, 30% sucrose solution for at least 5 d. Each brain was frozen and sliced (on a cryostat) into 60 μm coronal sections, mounted on slides, and stained with thionin. Subsequent examination of these sections was used to histologically verify the location of the electrode and location and extent of the IBO lesions.

Data analysis. All data were acquired with Datawave Discovery software (sampling rate for units, 22 kHz), and units were isolated using the Datawave Common Processing data analysis package. Position and head direction were sampled from the tracker at 6.25 Hz with linear interpolation being used to more accurately determine animal position for spike firing times occurring between any two of these samples. All data recorded while the tracker had mistracked (i.e., one or both LEDs ceased to be visible to the video camera) was disregarded. FR maps were constructed by dividing the floor of the chamber into a 40 × 20 grid in which each square (pixel) corresponded to 3 cm². For each pixel, the number of spikes occurring within it was divided by the amount of time spent in the same pixel (dwell time) to provide a firing rate. Unit data from a pixel where the animal spent <1 sec were considered undersampled and removed, leaving a firing rate map (Fig.1A) that was used for the calculation of: (1) the “mean FR”, the average firing rate of the unit, (2) the “mean in-field FR”, the mean firing rate inside the place field, (3) the “max in-field FR”, the maximum firing rate inside the place field, (4) the “mean out-field FR”, the mean firing rate outside the place field, and (5) “spatial discriminability”, the mean firing rate inside the place field divided by the mean firing rate outside the place field. To determine the half-amplitude PF size, the location of the peak of the place field and the number of fields, the data received minor smoothing (Fig.1B). This involved filling undersampled pixels in the firing rate maps with the average value of their neighboring pixels and then smoothing using a 3 × 3 normalized, equally weighted matrix. Place field pixels that were not adjacent to at least two other place field pixels were removed (Fig. 1C).

The stability of place field location was measured in three different ways: (1) Pearson's product-moment correlations (spatial correlation) calculated between each smoothed firing rate map were generated for each unit (both between and within sessions). (2) The shift in the location of the place field peak (peak shift) both between and within sessions was used as a second measure of stability and was determined by calculating the distance (in pixels) between the location of maximal firing in each smoothed place field map. The mean and SD of the distribution for units recorded from control animals were then calculated, and units (either from lesion animals or controls) that showed a peak shift of >2.33 SDs from this mean (i.e.,p < 0.01, one-tailed) were considered to exhibit significant instability in their place field locations. (3) The centroid (the average location of pixels in the field weighted by firing rate) (Fenton et al., 2000) and the subsequent centroid shift between and within sessions was calculated from the raw firing rate maps (with undersampling removed).

For within-session comparisons of peak shift, centroid shift, and spatial correlation, the firing rate map from the first half of the recording session was compared against that of the second half. Note that the peak shift, centroid shift, and spatial correlation were usually based on the average of six values for between-session data (i.e., every possible pairwise comparison of firing rate maps), and the average of four values for within session data (i.e., all sessions: baseline and three delays). All other measures were averaged over all sessions for that unit.

For statistical analyses, the type of t test used for pairwise comparisons was determined by a homogeneity of variance test. All t tests were two-tailed.

RESULTS

DISCUSSION

The stability of place cells recorded from region CA1 of the hippocampus was compared in control animals and animals with bilateral lesions centered on perirhinal cortex. It was determined that ∼36% of units recorded from animals with lesions of the perirhinal cortex had place fields that shifted location across a delay period. In contrast, no control place cells were unstable by this measure. Place cells in the two groups of animals did not, however, differ significantly on measures of within-session stability.

The most parsimonious explanation of this finding is that it is a direct result of the loss of input from the perirhinal cortex to the hippocampus. That is, although the perirhinal cortex is not necessary for the formation of the place field firing of hippocampal place cells, it contributes information necessary to the maintenance of their location specificity across a delay interval. Alternative accounts, for example, that it is a result of the small between-group difference in average session duration can be discarded on the basis of comparisons of short and long recording sessions described in the previous section. Furthermore, the possibility that the instability effects were an artifact of a reduction inspatial discriminability (lower in-field firing rates) in unstable units can also be ruled out because there were no significant differences on any measures (except those describing between-session stability) when these unstable units were compared with a group of control units with similarly low firing rates. Although we cannot completely rule out the possibility that the instability resulted from a movement of the electrode in lesioned (but somehow, not control) animals, we believe that this is unlikely as waveform amplitudes and waveshapes were highly stable across recording sessions.

Previous studies have shown that under normal conditions place fields can be stable for very long periods when an animal is repeatedly exposed to a familiar environment (Thompson and Best, 1990). In a novel environment, however, the location of the field can be completely unrelated to that in the familiar environment (Muller and Kubie, 1987). This change appears to result from the creation of a new representation of the novel environment (a “remapping”), which, once formed, is stable without affecting the integrity of existing maps (Bostock et al., 1991). The unstable units in the present study may, therefore, represent a remapping of the environment. It would be difficult to determine whether this is a “partial” (Knierim and McNaughton, 2001) or a “full” remapping without recording from a large number of place cells simultaneously. On the basis of the few cases in which several neurons were recorded at the same time in the present study, however, it is possible that both partial and full remapping may occur.

Remapping could result from a lesion-induced loss of mnemonic information regarding the animal's previous exposure to the recording chamber. In this situation, an animal might experience the chamber as a novel environment on each experimental session. The proposal that hippocampal place cell instability is a result of a loss of mnemonic information required to reconstruct a spatial representation is consistent with the results of previous behavioral studies in which it has been shown that the perirhinal cortex has an involvement in both spatial (Wiig and Bilkey, 1994a,b; Nagahara et al., 1995; Otto et al., 1997; Liu and Bilkey, 1998a,b,c; Murray et al., 1998; Wiig and Burwell, 1998; but see Glenn and Mumby, 1998; Aggleton and Brown, 1999) and object (Meunier et al., 1993; Mumby and Pinel, 1994; Wiig and Bilkey, 1995; Ennaceur et al., 1996: Murray, 1996; Buckley et al., 1997; Brown and Xiang, 1998) memory. This proposal is also consistent with a recent finding that cells in the perirhinal cortex exhibit place-specific responses (Burwell et al., 1998) that are modified by changes in the sensory environment. Furthermore, the perirhinal cortex has been shown to be involved in processing information related to the recognition of stimuli in multiple modalities (Otto and Eichenbaum, 1992; Suzuki et al., 1993; Young et al., 1997; Brown and Xiang, 1998) and the associations between stimuli (Bunsey and Eichenbaum, 1993; Higuchi and Miyashita, 1996; Buckley and Gaffan, 1998; Erickson and Desimone, 1999;Murray et al., 2000), suggesting that this region may have an important role in integrating object and spatial information (Murray et al., 1998) during memory processing.

Barnes et al. (1997) reported that aged rats displayed a similar instability in place field location in ∼30% of recording sessions conducted in a familiar environment but separated by a delay period of up to 1 hr. In contrast, young animals almost always showed stable place fields. It was suggested by these researchers that the instability of the place fields reflects the fact that the aged rats sometimes retrieve the incorrect cognitive map after a delay period because of an inability to recognize the environment as being familiar (Redish et al., 1998). Because perirhinal cortex is one of the earliest regions affected by neurofibrillary tangles in Alzheimer's disease and normal aging in humans (Arriagada et al., 1992; Price and Morris, 1999), it is of interest to speculate that this region is also damaged in aged rats. In this model, therefore, the deficits observed in theBarnes et al. (1997) study may be causally related to those observed in the present experiment.

Barnes et al. (1997) also suggested that place field instability may be because of the disruption of long-term potentiation (LTP) mechanisms in aged animals. This idea is consistent with findings that NMDA-dependent processes are important for maintaining the stability of place fields (Hargreaves et al., 1997; Kentros et al., 1998; for review, see Shapiro and Eichenbaum, 1999). Future studies could explore this possibility using interventions that selectively block LTP (Bilkey, 1996; Liu and Bilkey, 1996a,b; Cousens and Otto, 1998;Ziakopoulos et al., 1999) in the perirhinal cortex (PRC)-CA1 and/or PRC-entorhinal (Deacon et al., 1983; Burwell et al., 1995; Liu and Bilkey, 1996a, 1997; Burwell and Amaral, 1998a,b; Naber et al., 1999; Shi and Cassell, 1999) pathway. This could involve the intracerebral infusion of NMDA antagonists into these regions while simultaneously monitoring hippocampal place cell activity.

In previous behavioral studies it has been shown that animals with damage to perirhinal cortex have intact memory over periods of ∼0–60 sec but exhibit memory deficits for intervals of ∼2 min or longer (Liu and Bilkey, 1998a,c). Although we did not observe a delay-dependent instability effect in the present study, this may be attributable to the fact that we could not test animals with a delay shorter than 2 min. This limit resulted from the time required to disconnect the animal from the recording apparatus, wipe down the chamber, return the animal to its home cage for a moment, and then replace the animal in the apparatus. It is possible, therefore, that if a shorter delay had been used in the present study (e.g., removal for 30 sec), a delay-dependent effect may have been observed.

The lack of a delay-dependent effect raises the possibility that the observed instability of place field locations is not memory-related, but is a consequence of the removal of some sensory input to the hippocampus. A number of previous studies have, however, determined that minimal deficits in visual (Mumby and Pinel, 1994; Buffalo et al., 2000) or spatial discrimination (Liu and Bilkey, 1998a,b,c; Wiig and Burwell, 1998) occur at short delays after lesions of the perirhinal cortex, arguing against a perceptual deficit in these cases. In addition, visual input, which normally exerts strong control over place field locations, is not necessary for stable place field firing (Muller and Kubie, 1987; Quirk et al., 1990; Save et al., 1998, 2000).

An alternative non-mnemonic explanation is based on the finding that the repeated disorientation of animals before entry to an unchanged, familiar environment can result in a between-session instability in place field locations (Knierim et al., 1995; but see Dudchenko et al., 1997). Because disorientation was also used in the current study during the few cue rotation sessions, it is possible that this interacted with the perirhinal cortex damage to produce the present effect. This is, however, unlikely, because the disorientation procedure used by Knierim et al. (1995) was much more intensive (e.g., disorientation on every trial, disorientation both on the journey from the home cage and on the return, and placement into the cylindrical recording apparatus from a random location). Furthermore, in the present study, an unstable place field occurred in a lesion animal that had not experienced any previous disorientation episodes.

In summary, lesions of the perirhinal cortex may disrupt an animal's ability to reliably make use of cues from multiple modalities or to use the associations between these cues, for the purposes of recognizing the environment as a whole. The animal may, therefore, sometimes fail to correctly identify the environment as familiar after a delay, a process that may correlate with the hippocampus retrieving an incorrect spatial representation or inappropriately generating a new map for the familiar environment (remapping). This effect may explain recent findings that indicate that under some circumstances the perirhinal cortex contributes to spatial memory processes (Liu and Bilkey, 1998a,b,c; Murray et al., 1998; Sacchetti et al., 1999; Mumby and Glenn, 2000).

Footnotes

• This research was supported by grants from the Marsden Fund and University of Otago to D.K.B. Thanks also to Xiaodong Lu, Dr. Eric Hargreaves, Dr. Ping Liu, and Noah Russell for useful discussions and assistance.

  Correspondence should be addressed to Dr. David. K. Bilkey, Department of Psychology, University of Otago, P. O. Box 56, Dunedin, 9001, New Zealand. E-mail: sycodkb{at}otago.ac.nz.

  G. M. Muir's present address: Department of Psychological and Brain Sciences, Dartmouth College, Hanover, NH.