It is known that the entorhinal cortex plays a crucial role in spatial cognition in rodents. Neuroanatomical and electrophysiological data suggest that there is a functional distinction between 2 subregions within the entorhinal cortex, the medial entorhinal cortex (MEC), and the lateral entorhinal cortex (LEC). Rats with MEC or LEC lesions were trained in 2 navigation tasks requiring allothetic (water maze task) or idiothetic (path integration) information processing and 2-object exploration tasks allowing testing of spatial and nonspatial processing of intramaze objects. MEC lesions mildly affected place navigation in the water maze and produced a path integration deficit. They also altered the processing of spatial information in both exploration tasks while sparing the processing of nonspatial information. LEC lesions did not affect navigation abilities in both the water maze and the path integration tasks. They altered spatial and nonspatial processing in the object exploration task but not in the one-trial recognition task. Overall, these results indicate that the MEC is important for spatial processing and path integration. The LEC has some influence on both spatial and nonspatial processes, suggesting that the 2 kinds of information interact at the level of the EC.
It is well acknowledged that spatial cognition in mammals requires an interaction between the neocortex and the hippocampus. A key structure involved in this interaction is undoubtedly the entorhinal cortex (EC). With connections to and from a variety of cortical regions (reviewed by Kerr et al. 2007), the EC provides major input to the dentate gyrus, all subfields of the hippocampus and the subiculum via the perforant pathway. The EC has been subdivided, primarily on an anatomical basis, into 2 subregions, the medial entorhinal cortex (MEC) and the lateral entorhinal cortex (LEC), that innervate the hippocampal formation via the medial and the lateral perforant pathway, respectively. Although these 2 regions share similar intrinsic architecture, they display distinct patterns of extrinsic cortical and subcortical connections, as well as different electrophysiological and pharmacological properties, thus suggesting that the MEC and LEC mediate different functions (see Ferbinteanu et al. 1999; Sewards and Sewards 2003; Witter and Amaral 2004; Knierim 2006; Witter and Moser, 2006; Kerr et al. 2007; van Strien et al. 2009 for reviews). In particular, there is a consensus that the MEC and the LEC independently process spatial and nonspatial information, respectively (Zhu et al. 1995; Young et al. 1997; Naber et al. 2001; Hargreaves et al. 2005; McNaughton et al. 2006; Albasser et al. 2010; Yoganarasimha et al. 2010). This hypothesis has been recently supported by unit recording studies showing that the MEC contains neurons with spatial firing properties, including grid cells (Hafting et al. 2005), head direction cells, grid × head direction conjunctive cells (Sargolini et al. 2006), and border cells (Solstad et al. 2008), whereas neurons in the LEC show little spatial modulation (Hargreaves et al., 2005; Yoganarasimha et al., 2010).
Surprisingly, only a few lesion studies in rodents have examined the respective contribution of the MEC and the LEC to spatial cognition, and their results are not fully consistent with the proposed spatial/nonspatial dissociation. For example, Ferbinteanu et al. (1999) have shown that damage to the medial perforant path, but not the lateral perforant path, induced place-learning deficits during a water maze task, suggesting a role of the MEC in spatial memory. However, using a similar task, Burwell et al. (2004) found that performance was left unaffected by lesions including the MEC and the postrhinal cortex. Importantly, Hunsaker et al. (2007) have suggested that spatial and nonspatial processing may not be as clearly segregated when conveyed from the MEC and LEC to the hippocampus through the perforant path. Selective pharmacological inactivation of the medial perforant path (output to the dorsal hippocampus from the MEC) in rats performing an object exploration task mainly affected spatial processing whereas inactivation of the lateral perforant path (output to the dorsal hippocampus from the LEC) generally affected both nonspatial and spatial processing. A hypothesis consistent with this view was recently proposed by van Strien et al. (2009). This hypothesis is based on the anatomical observation that the projections of the postrhinal and perirhinal cortex to the EC overlap and that the MEC and the LEC are interconnected. As a result, spatial information in the MEC and nonspatial information in the LEC may be already associated at the level of the EC (van Strien et al., 2009). Such association would support the formation of episodic memory in the hippocampus.
We addressed the spatial/nonspatial hypothesis by directly comparing the effects of MEC and LEC lesions in a number of behavioral tasks that address different aspects of spatial and nonspatial information processing. Two exploration tasks allowed us to examine the specificity of MEC and LEC functions regarding spatial and nonspatial processing of intramaze objects. They included an object exploration task involving detection of spatial and nonspatial change (Save et al. 1992) and a one-trial recognition task involving spatial and object recognition (e.g., Langston and Wood 2010). We reexamined the involvement of the MEC and LEC in the Morris water maze, which emphasizes allothetic navigation. We then examined a path integration task that emphasizes idiothetic navigation. As grid cells in the MEC may implement a dynamically updated representation of the rat's location and orientation based on idiothetic cues (McNaughton et al. 2006), we hypothesized a path integration deficit in rats with MEC lesions but not rats with LEC lesions. Altogether, the results provide a general picture of the contribution of the MEC and the LEC in spatial behaviors and suggest an interaction between spatial and nonspatial processes at the EC level.
Materials and Methods
Male Long-Evans rats (Janvier, Le Genest-St-Isles, France) weighing between 275 and 300 g, housed in individual cages (40 cm long × 26 cm wide × 16 cm high) with ad lib food and water and kept in a temperature-controlled room (20° ± 2) with natural light/dark cycle. One week after arrival, animals were handled daily by the experimenter for 7 days. Prior to surgery, they were arbitrarily assigned to 3 groups: MEC-lesioned rats (MEC, n = 8), LEC-lesioned rats (LEC, n = 8), and sham-lesioned rats (SHAM, n = 8). The sham-lesioned group was composed of sham MEC–lesioned (n = 4) and sham LEC–lesioned rats (n = 4). As no differences were observed during the acquisition in all behavioral tasks, sham MEC and sham LEC rats were pooled in a single SHAM group. All rats from the 3 groups were used in the 3 experiments (navigation, object exploration, path integration). An additional set of rats (MEC, n = 8; LEC, n = 8; SHAM, n = 10) was used exclusively to examine path integration in rats that were trained before surgery. Following surgery, animals were allowed to recover for 10 days before testing was started or carried on.
The experiments were performed in accordance with the NIH guide for the care and use of laboratory animals (NIH publication no. 86-23, revised 1978), European guidelines (European Community Council Directive, November 2004, 1986, 86/609/EEC), and National guidelines (Council directive n°87848 of the Direction des Services Vétérinaires de la Santé et de la Protection Animale permission n° 13.24 from the Ministère de l’Agriculture et de la Pêche to E.S.).
Rats were deeply anesthetized by an intramuscular (i.m.) injection of xylazine (15 mg/kg; Rompun, Bayer, France) and ketamine (100 mg/kg; Imalgène, Merial, France) and placed in Kopf stereotaxic apparatus (Kopf instruments, Tujunga, CA). After a midline incision of the scalp was made, the skin and muscles were carefully retracted to expose the skull. Holes were drilled above the target regions. Bilateral lesions of the MEC or LEC were made by passing a radio-frequency current at the tip of an electrode (70 °C for 15 s; RFG 4, Radionics, Burlington, MA) lowered in the brain at the following coordinates relative to bregma: MEC (2 lesion points): AP: −8.3 mm, L: +4.8 mm; AP: −8.8 mm, L: +4.7 mm; LEC (2 lesion points): AP: −7.3 mm, L = 6.2 mm; AP: −6.6 mm, L: 6.4 mm (Paxinos and Watson 2004). For each lesion point, the electrode was lowered very slowly until its tip touched the floor of the brain or calvarium and then raised 1 mm (Parron and Save 2004a, 2004b). This position was taken as the dorsoventral coordinate of the lesion. Due to the anteroposterior curvature of the calvarium, this coordinate was different for each lesion point. Sham-operated rats were treated the same way as lesioned rats except that no current was passed through the electrode. After lesions, the skin was sutured and the rats received an injection of antibiotic (Terramycine, 60 mg/kg, i.m.; Pfizer, Paris, France) and analgesic (Tolfénide, 0.06 mg/kg, subcutaneous; Vetoquinol, Lure, France) as postoperative treatment. They were placed back in their home cage for recovery.
Apparatus and Behavioral Procedure
Water Maze Task (Experiment 1)
The water maze was an elevated circular pool (1.80 m diameter; 30 cm deep, 50 cm above the floor) located in a well-lit room containing numerous extra-maze cues (door, cupboards, shelves, pieces of equipment, posters, etc.) and filled with 20 ± 2°C water made opaque by addition of chalk powder. A white-painted, circular platform (diameter, 8 cm) was placed inside the pool, 30 cm away from the wall. Its top surface was 1 cm below the surface of the water and was therefore invisible to the animals. An overhead camera was connected to a computer and a DVD recorder. A radio set fixed to the ceiling in a central position produced background noise >70 dB to mask noncontrolled directional sounds. The rats were trained in a reference memory version of the navigation task in which the platform was held in a constant position throughout training. They received 4 daily trials for 6 days with a 30-s intertrial interval (rat left on the platform) and a 1-min trial limit after which the animal was gently guided toward the platform. A typical trial consisted of gently releasing a rat into the water with its head facing the wall, from 1 of 4 possible starting places (N, E, S, and W) around the perimeter of the pool. The 4 starting positions were used in a pseudorandom order within a 4-trial block. After completion of the 4 trials on the last training day, the rats received a 1-min no-platform probe trial.
Object Exploration Task and One-Trial Recognition Task (Experiments 2a and 2b)
The apparatus was a circular gray open field (100 cm diameter, 50 cm high), with a wooden floor that was painted gray. It was surrounded by a white curtain so that the environment was visually uniform and indirectly illuminated by four 40-W spotlights. An overhead camera was connected to a computer and a DVD recorder. As shown in Figure 1A, the open field contained 4 objects: (A) a Rubik’s cube (5.5 cm on each side) placed on a cylindrical metal base (total height, 22.5 cm), (B) a small glass bottle (13 cm high, 8 cm in diameter), (C) a plastic rectangular box (11 × 5 × 15 cm), (D) a dark gray plastic bottle (20 cm high, 8 cm in diameter). The novel object (E) was a white wood cylinder (21 cm high, 8 cm in diameter). Similar objects have been used in previous studies (e.g., Van Cauter et al. 2008a) and do not trigger any exploration preference bias. The distance between A, B, and C was approximately 30 cm, and D was located approximately 20 cm from the 3 other objects. From the spatial change session, all objects were at the same distance from each other and from the wall. Rats were submitted to 10 successive 4-min sessions with 4-min intersession intervals during which the animal was returned to its home cage (Save et al. 1992; Fig. 1A). In session 1, no object was placed in the arena, allowing the rat to familiarize with the environment. From sessions 2 to 7, the objects were placed in their standard configuration (Fig. 1A). In session 8, an object (D) was moved to a new position therefore yielding a different spatial configuration of the set of objects (spatial change). Session 9 was a rehabituation session similar to session 8. In session 10, a familiar object was replaced by a novel object with the spatial configuration held constant (nonspatial change). It has been repeatedly shown that control rats react to spatial and nonspatial change by selectively reexploring the displaced object and the novel object, respectively (e.g., Save et al. 1992; Lee et al. 2005; Hunsaker et al. 2007; Van Cauter et al. 2008a), indicating that the animals have processed and memorized the spatial and nonspatial characteristics of the environment.
In Experiment 2b (Fig. 1B), we aimed to test object and spatial recognition using a variant of the one-trial recognition task (Aggleton 1985; Ennaceur and Delacour 1988). We conducted this task to address the issue that deficits in the object exploration task may result from the complexity of the object configuration task in terms of spatial/nonspatial integration and memory load. Because there were only 2 objects, the one-trial recognition task involved less demanding object/place associations than the object configuration task. Basically, in this kind of task, rats are first exposed to 2 identical copies of an object (sample phase). After a delay, they are exposed to a novel object and a copy of the now familiar object (test phase). Control rats have been shown to explore more the novel object than the familiar object therefore indicating successful object recognition. In our experiment, we tested both object (a novel object replacing a familiar one) and spatial recognition (a copy of the familiar object in a novel place) (Dix and Aggleton 1999; Langston and Wood 2010). The apparatus was different from that used in Experiment 2a. It was a white circular arena (80 cm diameter, 50 cm high), surrounded by a circular curtain. A cue card (30 × 40 cm) was attached to the arena wall to avoid ambiguity in object location when identical objects were used. The 2 objects were at a 25-cm distance from each other and from the cue card. The experiment was performed across 3 consecutive days (Fig. 1B). On day 1, rats were submitted to a 10-min familiarization session in the empty arena. On day 2, rats were tested for object recognition. In a presentation session (duration 2–5 min or until each object have been explored 15 s), they were exposed to 2 identical objects. After a 2-min delay during which the animal was put back in its home cage, a 3-min test session was performed. In this test, 1 of the 2 familiar objects was replaced by a novel object (with balanced right/left substitution). Preferred exploration of the novel object would indicate novel object recognition. On day 3, rats were tested for spatial recognition. In a presentation session (duration 2–5 min or until each object had been explored 15 s), they were exposed to a single object (different from the objects used on day 2, balanced right/left presentation). After a 2-min delay, a 3-min test was performed during which the rat was exposed to 2 objects similar to the object of the presentation session. In both experiments 2a and 2b, accurate spatial recognition was indicated by preferred exploration of the object that occupied a new location.
Path Integration Task (Experiment 3)
The apparatus was a white circular elevated platform (190 cm diameter, 85 cm above the floor) located in a well-lit room (see Save et al. 2001; Parron and Save 2004b for similar apparatus). Eight holes (12 cm diameter) were evenly distributed along the periphery and served as starting places for the animal. Seventeen gray polyvinyl chloride cups that could each contain a piece of food pellet were located on the central area of the platform (Fig. 1C). The animal was brought from the colony room in a transportation box containing sawdust. This “home” box was placed under 1 of the 8 peripheral holes by means of sliding guides. Environmental cues were carefully controlled. To eliminate remote visual cues, the apparatus was surrounded by a white opaque circular curtain. To mask auditory directional cues, a radio set fixed to the ceiling in a central position produced background noise >70 dB. To neutralize directional olfactory cues, 7 boxes similar to the transportation box, also containing soiled sawdust, were placed under each of the remaining peripheral holes. In addition, the platform was entirely cleaned with a sponge impregnated with 95% alcohol between each trial. More generally, to prevent the animal from relying on remote cues in the room reference frame, the starting hole and baited cup were different in each trial. The animals were submitted to a training phase and a test phase. The animals were progressively trained so that at the end of training, they readily jumped through the hole on the platform, explored it until they find the baited cup, took the pellet and carried it back directly to their starting hole. Training included 2 phases. In the first phase, all 17 cups were baited. Rats were submitted to several trials a day until they correctly produced the expected behavior on the first trial in 3 consecutive days. In the second phase, only one cup among the 17 was baited. The procedure and training criterion were similar as in the first phase. Following training, the animals were submitted to 34 test trials (2 for each cup). For each trial, the starting hole and the baited cup was changed until each cup has been visited once.
Water Maze Task (Experiment 1)
A videotracking software (Videotrack 1.7; View Point, Champagne-au-Mont-d'Or, France) was used to record and analyze the animals' behavior (sampling: 40 ms). In all trials, the trajectory length and swimming speed were measured. In the probe trial, we measured the time spent by the rat in the 4 quadrants and, to estimate precise platform searching, in 20-cm circular areas centered on the actual platform location (in the middle of the target quadrant) and on an equivalent position in the 3 other quadrants.
Object Exploration and One-Trial Recognition Tasks (Experiments 2a and 2b)
All sessions were recorded on DVD and analyzed off-line. As in previous studies (Parron and Save 2004a; Van Cauter et al. 2008a), we measured the time a rat spent exploring each object. Exploring an object was defined as directing the snout at a distance of less than 2 cm of an object and actively exploring it (a “contact”). The duration of contacts was measured with a stopwatch by a single trained person who was blind to the different groups. Locomotor activity was measured by using the videotrack system. To evaluate the exploratory reaction to spatial change, we calculated a reexploration score for the displaced object and the nondisplaced objects that was the averaged time spent exploring the displaced and nondisplaced objects (average time for the non displaced objects) in session 8 minus session 7. Similarly, to evaluate the reaction to nonspatial change, we calculated a reexploration score for the novel object that was the time spent exploring the novel object minus the averaged time spent exploring the familiar objects in session 10.
In the one-trial recognition task, we measured the time spent exploring each object during the first, second, and third minutes of the test. Consistent with the results of Dix and Aggleton (1999), rats mostly explored the objects during the first 2 min. A discrimination score (Langston and Wood 2010) was calculated for each rat for the first 2 min using the formula (time at novel − time at familiar)/(time at novel + time at familiar) where novel refers to the novel object or the object at the novel location.
Path Integration Task (Experiment 3)
All sessions were recorded on DVD and analyzed off-line. We used a number of parameters including the number of correct responses (direct return to starting place) and complexity of the outward path. Path complexity was estimated by calculating the ratio between the actual distance run by the animal to reach the rewarded cup from the starting hole and the shortest distance between these 2 places.
At the end of the experiment, all rats were sacrificed with an overdosed lethal sodium pentobarbital and perfused transcardially with 0.9% NaCl followed by 4% paraformaldehyde solution. The brains were embedded in paraffin to preserve the delicate structure of the lesion-damaged brains, before 30-μm coronal sections were cut on a microtome. Every second section was mounted and stained with cresyl violet. Sections were studied with a Zeiss microscope with a mounted camera. All pictures were processed with AxioVisionLE (Carl Zeiss Vision, GmbH). For MEC rats that had been trained in the path integration task only post lesion, we calculated lesion extent to assess its relationship with their impaired performance on that task. We manually traced the outline of tissue loss on each section and multiplied these areas by the intersection distances (see Results).
Figure 2 provides a schematic representation of MEC (left panel) and LEC (right panel) lesions, including rats that underwent postlesion training (all behavioral tests) and those that were submitted to prelesion training (path integration). The minimum and maximum lesion extents are represented in dark and light gray, respectively. There was clear tissue loss in the target entorhinal areas and no extensive damage of the hippocampus, the presubiculum, and the perirhinal and postrhinal cortices. In MEC rats, lesions slightly encroached upon the parasubiculum, particularly at more posterior lesion sites, but in no case did they damage the LEC. In LEC rats, the largest lesions slightly encroached upon the MEC. Such damage was anterior to the region where grid cells are classically recorded. In both MEC and LEC groups, lesions extended from the dorsolateral to the ventromedial bands (Steffenach et al. 2005). The lesion extended largely to the ventral portion of the MEC in 6 out of 8 MEC rats and partially in the 2 remaining rats.
Deficits were obtained following radio-frequency lesions that do not spare fibers of passage. The question of whether the behavioral deficits could result from such damage and not from or in addition to EC damage is raised. As repeatedly described in the literature, the vast majority of the cortical inputs to the parahippocampal region displays a synaptic relay in the MEC and LEC (e.g., Witter 2007). There are some direct projections from the perirhinal and postrhinal cortices to the CA1/subiculum, but they do not run through the EC (Naber et al. 1999, 2001). Thus, the amount of putative fibers of passage may be quite modest. Should they be damaged, the behavioral impact would be much weaker than the impact of EC damage. Thus, it is likely that the behavioral deficits found in the present study are a consequence of EC damage rather than lesions of fibers of passage.
Experiment 1: Water Maze Navigation Task
MEC rats swam at a greater swimming speed than both SHAM and LEC rats (one-way analysis of variance [ANOVA]: F2,21 = 56.05, P < 0.001; Newman–Keuls post hoc tests: MEC > SHAM, MEC > LEC, SHAM = LEC, all P < 0.001). In spite of this, as shown in Figure 3A, they swam longer distances to reach the platform (2-way ANOVA: effect of lesion, F2,21 = 11.62, P < 0.001; Newman–Keuls: MEC vs. SHAM, P < 0.001, MEC vs. LEC, P < 0.01). However, all 3 groups improved over sessions (2-way ANOVA: session effect, F5,105 = 12.58, P < 0.001, no session × group interaction, F10,105 = 0.85). There was no statistically significant difference at the end of training (S6, one-way ANOVA: F2,21 = 3.04, P = 0.07).
As shown in Figure 3B, during the probe test, all groups displayed a preference for the target quadrant (2-way ANOVA: no group effect, F2,42 = 2.00, P = 0.15, quadrant effect, F1,42 = 7.99, P < 0.01) and furthermore were able to concentrate their search on the platform location (2-way ANOVA: no group effect, F2,42 = 2.12, P = 0.13, area effect, F1,42 = 87.10, P < 0.001). However, MEC rats showed lower searching behavior (goal area, group × area interaction, F2,42 = 5.40, P < 0.01 and Newman–Keuls: MEC < SHAM, P < 0.01; MEC < LEC, P < 0.025, LEC vs. SHAM, P > 0.05; quadrant, group × area interaction, F2,42 = 7.99, P < 0.01 and Newman–Keuls: MEC < SHAM, P < 0.001, MEC < LEC, P < 0.01, LEC vs. SHAM, P > 0.05).
Overall, MEC rats were impaired relative to SHAM and LEC rats in learning the platform location as they swam longer distances in spite of faster swimming speed. In addition, MEC rats showed a preference for the platform location during the probe test, but their preference was less precise than that of SHAM and LEC groups.
Experiments 2a: Object Exploration Task
Figure 4A,B show the time course of locomotor activity and object exploration activity across sessions in the 3 groups, respectively. All groups displayed habituation of both locomotor activity (repeated-measures ANOVA: session effect, F6,126 = 62.75, P < 0.001), although at different rate (group × session interaction, F12,126 = 4.22, P < 0.001) and object exploration (repeated-measures ANOVA: no group effect, F2,21 = 1.08, P = 0.35, session effect, F5,105 = 6.21, P < 0.001; no group × session interaction, F10,105 = 1.52, P = 0.14). MEC rats showed hyperlocomotion, and LEC rats were less active than SHAM rats (group effect, F2,21 = 19.50, P < 0.001 and Newman–Keuls: MEC > SHAM > LEC, P < 0.001 for all comparisons), but all groups displayed similar level of locomotor activity (Newman–Keuls, all comparisons P > 0.05) and object exploration (one-way ANOVA, F2,21 = 0.92) at the end of habituation (session 7).
As shown in Figure 4C, the reexploration score of displaced and nondisplaced objects significantly differs between SHAM and both MEC and LEC rats (2-way ANOVA: group effect, F2,42 = 3.79, P < 0.05, object (displaced vs. nondisplaced) effect, F1,42 = 9.64, P < 0.01, group × object interaction close to significance, F2,42 = 3.14, P = 0.054. SHAM rats explored the displaced object significantly more that the nondisplaced objects (Newman–Keuls, P < 0.01). This was not the case for MEC and LEC rats (P = 0.436 and P = 0.998, respectively). In addition, SHAM rats reexplored the displaced object more than both MEC and LEC rats (P < 0.05 and P < 0.01, respectively). This difference was not observed for the nondisplaced objects (all comparisons P > 0.05). Thus, MEC and LEC rats were not able to detect the spatial change, suggesting a role of the 2 structures in the processing of object location.
SHAM and MEC rats but not LEC rats showed a clear preference for the novel object (Fig. 4D, one-way ANOVA: F2,21 = 5.07, P < 0.05, Newman–Keuls: SHAM vs. LEC, P < 0.05; MEC vs. LEC, P < 0.05, SHAM vs. MEC, P > 0.05). Thus, only LEC rats were impaired in the detection of the novel object. This deficit could not be ascribed to a generalized decrease in exploratory activity since both the duration and the number of contacts with objects across sessions 2 to 10 were similar between LEC and SHAM animals (number of contacts, repeated-measures ANOVA: group effect, F2,21 = 4.62, P < 0.05, session effect, F8,168 = 6.12, P < 0.001, no session × group interaction, F16,168 = 0.76, and Newman–Keuls: SHAM vs. LEC, P > 0.05), whereas MEC rats made more contacts than both SHAM and LEC rats (both P < 0.05).
Experiment 2b: One-Trial Recognition Task
Experiment 2b examined whether spatial deficit obtained in LEC rats was dependent on the complexity of object/place processing by using a one-trial recognition task involving only 2 objects instead of 4.
As shown on Figure 5, the discrimination score for object recognition was not different among the 3 groups (one-way ANOVA: F2,21 = 0.55), indicating that all groups were able to detect the novel object. In contrast, the discrimination score for spatial recognition in the MEC group was significantly different from both SHAM and LEC animals (one-way ANOVA: F2,21 = 10.38, P < 0.001, Newman–Keuls: MEC vs. SHAM, P < 0.001, MEC vs. LEC, P < 0.01). SHAM and LEC rats discriminated between the 2 objects with a preference for the object at the new location whereas MEC rats did not (one-sample t-tests—comparison of discrimination score to chance performance, SHAM: t7 = 5.64, P < 0.001; LEC: t7 = 4.10, P < 0.01; MEC: t7 = 1.30, P > 0.05).
Overall, the results of experiments 2a and 2b show that MEC rats were impaired in the processing of spatial information but not in the processing of nonspatial information. In contrast, LEC rats were impaired in spatial and nonspatial processing if the task involved complex object/place processing but not impaired if the task was more simple.
Experiment 3: Path Integration
All the rats had been submitted to the water maze navigation and object exploration, and one-trial recognition tasks were also trained in the path integration task (postlesion training). To rule out any lesion effect on acquisition of the basic procedural rules of the task, we included an additional group of rats that was trained prior to receiving lesions (prelesion training).
Rats were trained until they exhibited appropriate return behavior on the first trial in 3 consecutive days (see Apparatus and Behavioral Procedure). The mean number of days to reach the training criterion was 11.6 for SHAM rats, 13.5 for MEC rats, and 12.0 for LEC rats in the postlesion training condition and 11.1 for SHAM rats, 9.9 for MEC rats, and 11.9 for LEC rats in the prelesion condition. Groups with postlesion and prelesion training learned the procedural aspects of the task at similar rate (one-way ANOVA: postlesion training, F2,21 = 0.49, prelesion training, F2,23 = 1.28, P = 0.30).
The number of correct returns in groups that combined rats with prelesion and postlesion training is shown in Figure 6. MEC rats showed a clear deficit whereas LEC rats displayed sham-like performance (one-way ANOVA: F2,47 = 5.26, P < 0.01, Newman–Keuls: MEC < SHAM, P < 0.05, MEC < LEC, P < 0.01, LEC vs. SHAM, P > 0.05). Thus, MEC rats but not LEC rats exhibited a path integration deficit.
The path integration deficit of MEC rats was not a consequence of more complex outward paths as the path complexity index was similar among the 3 groups (SHAM: 4.21, MEC: 3.83, LEC: 3.73, combining postlesion and prelesion training conditions) (one-way ANOVA: F2,1697 = 2.48, P = 0.08).
In order to examine whether path integration deficits depended upon the extend of MEC lesion, we calculated the area of tissue damaged for each MEC rat in the postlesion–trained group. The extent of MEC lesion ranged from 0.42 to 1.39 mm3 (mean = 0.72 ± 0.06 mm3). There was no significant correlation between MEC lesion size number of correct returns (Pearson product–moment and Spearman correlation coefficients, r = 0.35; P > 0.05; r = 0.34; P > 0.05, respectively).
Overall, these results indicate that MEC rats showed a real deficit in path integration, that is, that was not secondary to an impairment in procedural training.
The functional distinction between the MEC and LEC has been at best controversial. Only a handful of studies have examined the behavioral effects of selective lesions of these 2 regions (e.g., Ferbinteanu et al. 1999; Burwell et al. 2004; Hunsaker et al. 2007), yielding mixed results and thereby allowing limited conclusions. In the present study, we trained MEC- and LEC-lesioned rats in several tasks requiring different kinds of spatial and nonspatial processing. This provided a more complete picture of involvement of the MEC and LEC in such processing. We found that rats with MEC lesions were impaired in the processing of spatial information in both the object exploration and one-trial recognition tasks, but nonspatial object recognition was unimpaired. MEC lesions mildly affected place navigation in the water maze and produced a path integration deficit. In contrast, LEC lesions spared navigation abilities in both the water maze and the path integration tasks. Spatial and nonspatial processing of intramaze objects were markedly reduced in the object exploration task but neither was affected in the one-trial recognition task.
The deficits observed following MEC lesions suggest an involvement of this structure in various aspects of spatial information processing, that is, spatial novelty detection and path integration. Indeed, MEC-lesioned rats were not able to detect or recognize a spatial change (novel spatial arrangement of familiar objects) but detected the nonspatial change (a novel object) in tasks involving spontaneous object exploration. The deficit was specifically due to the inability to process spatial information and could not be ascribed to a general memory deficit, which would have impaired the detection of both spatial and nonspatial novelty. Similarly, the possibility that exacerbated locomotor activity of MEC rats affected object exploration can be ruled out as these animals showed control-like object exploratory activity in terms of number and duration of contacts with the objects across all sessions. The effects of MEC lesions are consistent with the function generally attributed to the postrhinal–MEC circuit, conveying visuospatial inputs and polymodal inputs from posterior association regions (e.g., parietal, retrosplenial cortex). Unimodal inputs and inputs from anterior associational regions (frontal, insular cortex) are conveyed through the perirhinal–LEC circuit (Burwell 2000).
MEC lesions mildly affected performance in the water maze task in which rats learn to locate a hidden platform using allothetic distal cues. This behavioral effect is consistent with previous unit-recording work showing that extended (Van Cauter et al. 2008b) or selective (Brun et al. 2008) entorhinal lesions did not abolish location-specific firing of hippocampal place cells, thought to be important for allothetic spatial navigation. However, navigation ability was more impaired when rats had to rely on idiothetic information, that is, during path integration. As rats were tested in both water maze and path integration tasks, it is difficult to interpret the overall pattern of results in terms of residual MEC activity from the tissue spared by the lesion. In addition, the histological analysis revealed that lesions extended to different domains of the MEC, including the lateral, intermediate, and medial bands, that may convey and process different spatial and nonspatial signals (Witter et al. 1989; Dolorfo and Amaral 1998; Fyhn et al. 2004; Kerr et al. 2007). Thus, our results point to the MEC as an essential part of a path integration-dependent navigation system, a finding consistent with the firing properties of MEC neurons (McNaughton et al. 2006). Indeed, the dorsomedial part of the MEC contains specific neurons, including grid cells, head direction cells, and conjunctive cells, whose firing activity may allow continuous updating of the animal self localization through integration of information about position, direction, and speed, during exploration (Hafting et al. 2005; Sargolini et al. 2006). The present results suggest that the path integration deficits originated from damage to the grid cell system. In addition, no correlation between lesion size and impairment in path integration was found, indicating that even a small perturbation of this network induced a large deficit. This finding is consistent with the hypothesis that the grid cell system is part of a network involved in computations necessary for path integration. Damage to this network may also explain less precise swimming paths during water maze navigation but spared memory of the platform location during the probe trial.
Our results revealed a pattern of behavioral deficits in LEC rats that differs from MEC rats. LEC-lesioned animals were not impaired in the water maze navigation task or in the path integration task. In the object exploration task, LEC rats were impaired in detecting both spatial and nonspatial changes, therefore suggesting a role of the LEC in the processing of spatial and nonspatial information of intramaze objects. Surprisingly, no deficit was found in the one-trial (spatial and object) recognition task. Thus, although the 2 tasks share basic aspects and exploit a similar behavioral repertoire, they are not equivalent in terms of information processing requirements—the one-trial recognition task being to some extent simpler than the object exploration task in terms of spatial/non spatial processing. One possible explanation is that LEC rats were able to efficiently use the cue card attached on the wall to discriminate object locations in the one-trial recognition task but not in the object exploration task. However, the presence of a similar cue card in the object exploration task was not sufficient to overcome the spatial deficit of LEC rats possibly due to the increased number of objects. It is relevant to discuss the deficits of LEC rats in terms of dysfunction of the perirhinal cortex–LEC–hippocampus (PER-LEC-HIP) interaction. In a situation involving numerous objects, the activity of this pathway may be important for conjoint nonspatial and spatial processing, allowing formation of multiple associations between objects and places in addition to object recognition (Bussey et al. 2001; Mumby et al. 2002; Winters et al. 2004; Barker et al. 2007; Warburton and Brown 2010). Spatial information may be used to disambiguate object identity (Brown et al. 2010; Jo and Lee 2010, for a review). Spatial/nonspatial conjunctive processing may occur in the EC itself or downstream in the hippocampus as inactivation of the lateral perforant path at this level affects both spatial and nonspatial processing (Hunsaker et al. 2007). When simpler object discrimination is required, the integration of spatial and nonspatial transmission may be less critical, and LEC dysfunctions would not result in object recognition deficits. In that case, transmission of both information types would still be possible by alternative pathways. Spatial information could be conveyed by the postrhinal cortex–MEC circuit and nonspatial information may reach the hippocampus via the direct perirhinal cortex–hippocampus projections (Naber et al. 1999). Thus, the overall scenario implies that the LEC is part of a circuit, the PER-LEC-HIP pathway, that contributes to combining spatial and nonspatial information but that its influence is dependent on environmental complexity.
In conclusion, while our results still comply with the hypothesis that the MEC plays a role in the processing of spatial information and the LEC in the processing of nonspatial information, they suggest that the 2 kinds of processing closely interact and eventually converge within the EC. In particular, the LEC would be involved in the conjoint processing of spatial and nonspatial information. Such interaction seems to be moderated by the task requirements and may involve an interplay between the PER-LEC and the postrhinal cortex–MEC circuits (see also Hunsaker et al. 2007 for a similar view). These results open the question of whether MEC and LEC functions are totally segregated or whether the degree of integration and interaction between these 2 areas is greater than observed in the upstream structures (PER and POR). The interaction between the 2 entorhinal structures may occur through anatomical interconnections, as recently suggested by van Strien et al. (2009) or through modulation of downstream target areas, such as the hippocampus. In addition, we provide evidence that the MEC is also involved in the processing of idiothetic information, a finding compatible with recent models of EC functions based on the existence of grid cells (McNaughton et al. 2006). Thus, our results and others stress the importance of dissecting not only the respective contribution of the MEC and LEC but also the intrinsic functional organization in these 2 regions to understand the function of the whole EC and the entorhinal–hippocampal interactions.
Centre National de la Recherche Scientifique; Ministère de la Recherche.
We thank Henriette Lucchessi for help with data analysis, Claire Maréchal for histological work, the Spatial Cognition group for discussion, Bruno Poucet, Pierre-Pascal Lenck-Santini, and Murray Horne for helpful suggestions. Conflict of Interest : None declared.