A 5-Month Period of Epilepsy Impairs Spatial Memory, Decreases Anxiety, but Spares Object Recognition in the Lithium-pilocarpine Model in Adult Rats

Authors


Address correspondence and reprint requests to Dr. J. Detour at INSERM U666, Clinique Psychiatrique, Hôpitaux Universitaires, 1 place de l'hôpital, BP426, 67091 Strasbourg Cedex, France. E-mail: Julien.Detour@chru-strasbourg.fr

Abstract

Summary: Purpose: In temporal lobe epilepsy (TLE), interictal behavioral disorders affect patients' quality of life. Therefore we studied long-term behavioral impairments in the lithium-pilocarpine (li-pilo) model of TLE.

Methods: Eleven li-pilo adult rats exhibiting spontaneous recurrent seizures (SRSs) during 5 months were compared with 11 li-saline rats. Spatial working memory was tested in a radial arm maze (RAM), anxiety in an elevated plus-maze (EPM), and nonspatial working memory in an object-recognition paradigm. Neuronal loss was assessed on thionine brain sections after behavioral testing.

Results: In the RAM, the time to complete each session and the number of errors per session decreased over a 5-day period in li-saline rats but remained constant and significantly higher in li-pilo rats. In the EPM, the number of entries in and time spent on open arms were significantly higher in li-pilo than li-saline rats. In the object-recognition task, the two groups exhibited a comparable novelty preference for the new object. Neuronal loss reached 47–90% in hilus, CA1, amygdala, and piriform and entorhinal cortex.

Conclusions: In li-pilo rats having experienced SRS for 5 months, performance in the object-recognition task is spared, which suggests that object discrimination remains relatively intact despite extensive damage. Neuronal loss in regions mediating memory and anxiety, such as hippocampus, entorhinal cortex, and amygdala, may relate to impaired spatial orientation and decreased anxiety.

It is generally agreed that patients with temporal lobe epilepsy (TLE) are more prone to behavioral disorders and cognitive impairments than is the general population (1). These interictal impairments usually disrupt the patient's everyday life, often more than the seizures themselves (2,3). Furthermore, depression or anxiety often develops several years after epilepsy onset (2,4), and the control of the seizures with pharmacologic treatment does not always suppress these disorders (5). Among cognitive impairments, memory problems are frequently observed in patients with TLE. The most reliable observations are deficits in declarative memory (ability to acquire facts and events related to one's personal past, 6) and in the performance of visuospatial tasks (7–9). Moreover, long duration of refractory TLE seems to be associated with cognitive deterioration (10–12). Thus understanding the neural mechanism underlying these disturbances is an important issue in the management of TLE. In animal models of TLE, the molecular, lesional, and metabolic characteristics have been quite extensively studied. Conversely, the cognitive or behavioral validity of the models was less studied, especially after a long-lasting period of spontaneous seizures.

The model of epilepsy induced in rats by pilocarpine alone or associated with lithium reproduces most clinical and neuropathologic features of human TLE (13–17). In adult rats, the injection of lithium and pilocarpine (li-pilo) leads to status epilepticus (SE) followed by a latent seizure-free period of a mean duration of ∼3 weeks, after which all animals exhibit spontaneous recurrent seizures (SRSs) that last for their whole life (chronic period). During the latent period, neuronal loss, mossy fiber sprouting, gliosis, and synaptic reorganization participate in the constitution of a hyperexcitable circuit that underlies the occurrence of SRSs. Neuronal loss is located mainly in hippocampus, parahippocampal cortices, amygdala, and thalamus. Most of these structures are involved in behaviors like memory and anxiety (18,19). The behavioral consequences of TLE cannot be solved by the traditional approach that usually disrupts one or two regions of interest at the most. However, the lesions induced by li-pilo SE that represent multifocal brain damage resembling the neuropathology of human TLE can be used as an adequate tool for the understanding of interictal behavioral disorders in patients with TLE (17).

Previous studies on the behavioral consequences of SRSs consecutive to pilocarpine- or kainate-induced SE have concentrated mostly on memory. Thus working memory appears impaired in the Morris water maze during the latent period (20), and spatial working memory assessed in a radial arm maze is impaired during the chronic period (17,21–24). In the present work, we evaluated the long-term behavioral consequences of li-pilo SE induced in adult rats. The main difference with previous studies is that behavioral performance was assessed after a quite long period of epilepsy (i.e., 5 months of SRSs). To evaluate spatial working memory, we used the eight-arm maze (25). Deleterious effects of hippocampal lesions in learning and spatial memory have been reported in this task (18,26,27). We also used two other tasks that involve behaviors mediated by structures quite vulnerable to li-pilo SE, like the parahippocampal cortices or amygdala: the elevated-plus maze and an object-recognition task. The elevated-plus maze is designed to assess the level of anxiety in rodents (28), which depends partly on the amygdala (19). The last test was a nonspatial object-recognition task. Recognition memory is generally regarded as the ability to discriminate the familiarity of things previously encountered (29), which is critical for everyday life. This behavior, which seems to depend mostly on structures outside of the hippocampus, mainly perirhinal and entorhinal cortices (for review, see refs. 29 and 30), has never been explored in the li-pilo model. In addition, the extent of neuronal loss was assessed in the regions of interest of the same rats.

MATERIALS AND METHODS

Animals

In total, 58 adult Sprague–Dawley rats (Janvier Breeding Center, Le Genest-St-Isle, France) weighing 300–350 g were used for the study. All rats where housed in quiet, uncrowded facilities in a room maintained at 21–22°C on a 12/12-h light/dark cycle (lights on at 7.00 a.m.) with food and water available ad libitum. Rats were maintained in groups of four animals before the pilocarpine protocol and were then housed individually for the remainder of the experiment. Males only were used for the experiments to eliminate confounding effects of variable estrogen levels on neuronal excitability. All animal experimentation was performed in accordance with the rules of EC Council Directive (86/69/EEC) of November 24, 1986, and the French Department of Agriculture (license no. 67–97). All efforts were made to minimize animal suffering.

Pilocarpine protocol

Animals were randomly divided into an experimental (li-pilo group, n = 41) and a control group (li-saline group, n = 17). All rats received lithium chloride (3 mEq/kg; Sigma, St. Louis, MO, U.S.A.) intraperitoneally. Then, 20 h later, the li-pilo group received 1 mg/kg methylscopolamine bromide (s.c., Sigma) to limit the peripheral effects of the convulsant. SE was induced 30 min later by the subcutaneous injection of pilocarpine hydrochloride (25 mg/kg, Sigma). Diazepam (2.5 mg/kg; Valium, Roche, Meylan, France) was injected i.m. 2 h after SE induction and 4 h later to improve survival. The li-saline group received methylscopolamine and saline instead of pilocarpine. All animals were aged 90 days at the time of SE induction.

The occurrence of SRSs was observed in the li-pilo group starting from day 14 after the pilocarpine injection. The observation period was conducted daily from 9.00 a.m. to 6.00 p.m. All rats were observed until the occurrence of at least one seizure.

Behavioral testing

Among the 21 epileptic rats that survived SE, the 11 li-pilo rats that exhibited the most frequent SRSs were selected for behavioral studies and compared with 11 li-saline rats randomly selected within the 17 available. Behavioral performance and neuronal loss were assessed at ∼5 months after SE. To test anxiety, spatial working memory, and novelty discrimination, all rats were tested in the elevated-plus maze, the eight-arm maze, and an object-recognition task, respectively, over a period of ∼1 month. Three weeks before the first behavioral testing, rats were housed under an inverse 12/12h dark/light cycle (lights off at 7.00 a.m.). All behavioral testing was performed ≥3 h after the last motor seizure, and no spontaneous seizure occurred in any rat during behavioral testing.

The elevated-plus maze

The elevated-plus maze is a validated test that evaluates anxiety in rodents (28). The apparatus was totally made of transparent Plexiglas. It comprises two open arms (50 × 10 cm), two enclosed arms (10 × 40 × 50 cm), and a central platform (10 × 10 cm). The configuration has the shape of a plus sign, and the apparatus is elevated 50 cm above the floor level. Grip on the open arms is facilitated by inclusion of a small edge (0.5 cm high) around their perimeter. For testing, rats were brought to the room 2 h before the test and were tested individually. Before each trial, the maze was cleaned thoroughly with a 30% ethanol solution. At the beginning of the test, rats were placed on the central platform always facing the same open arm. The test lasted 5 min in standard laboratory conditions under red light (2 × 60 W). The testing device was video-recorded, and the experimenter supervised the test in an adjacent room. Videotapes were scored by a naïve trained observer using a software developed in our laboratory. Behaviors were encoded afterward directly on a PC keyboard. Data were then transferred for statistical analysis.

Activity and anxiety-related behaviors were assessed. Standard measures comprised: the total number of arm entries (arm entry defined as all four paws entering an arm), the number of open- and closed-arm entries, and the time spent in different sections of the maze (open and closed arms and central platform). In addition to conventional measures, three specific behavioral measures were recorded: rearing frequency and duration, head-dipping frequency (exploratory movement of head/shoulders over the sides of the maze), grooming frequency and duration (typical sequences beginning with snout, progressing to ears, and ending with whole-body groom).

The eight-arm maze

Rats were tested in the eight-arm maze according to the procedure originally proposed by Olton and Samuelson (25) and described elsewhere (31). The animals were food deprived to 85% of their body weight before testing and maintained at that weight during the 5 days of testing. The apparatus was a wooden gray, enclosed eight-arm radial maze with walls and entirely covered with a transparent top. Each of the arms (60 × 12 × 17 cm) projected from one side of an octagonal center measuring 50 cm in diameter. One food pellet (45 mg) was positioned at the far end of each arm. Before each trial, every arm of the maze was baited with a food pellet. Reinforcement was not replaced during the test. The whole apparatus was video-recorded from above under red light (2 × 60 W). At the beginning of the test, a white plastic cylinder (45 cm in diameter) was used to place the rats in the central platform. The cylinder was taken back, and animals were left in the maze until they had either entered all eight arms or until 15 min had elapsed, whichever occurred first. Placing all four paws inside an arm was recorded as an arm entrance. Times of arm entrances, the identity of each arm entered, and the serial order were recorded. All rats were tested for 5 days (one test per day).

The object-recognition task

Apparatus The object-recognition task is a nonrewarded paradigm based on the spontaneous exploratory behavior of rats, which implies working memory. The test apparatus consisted of a fenced cage (50 × 40 × 30 cm) with a front door. Three objects were fixed on one side (25 cm above the floor) and could not be displaced by the rats. Distance between objects and walls was equal. Two of the three objects were made up of wood and were circular (diameter, 40 mm). The third one was made of plastic and had roughly the same aspect but was denticulate and smaller (diameter, 25 mm). The two types of objects had the same color. The objects had no genuine significance for rats and had never been associated with reinforcement. A video camera was mounted opposite to the cage and was used to record performance.

Procedure The object-recognition task was performed as follows. On the first day, rats were placed in the cage and were allowed to move freely for 5 min; the three objects were identical. Rats were then put back to their home cage. The same procedure was repeated on 3 consecutive days (one session per day). On the fourth day, the familiar object on the right was removed and replaced by the new denticulate object. This paradigm was chosen because of expected working-memory impairment in the li-pilo group (17) and to allow a habituation period (the environment of the fenced cage) before the recognition-testing day. Thus the intertrial interval was 24 h. To avoid the presence of olfactory trails, the apparatus and the objects were thoroughly cleaned with a 30% ethanol solution.

Video tapes were scored afterward by a naïve trained observer. Behaviors were encoded directly on a PC keyboard. The screen was divided into three zones, each zone being centered on one object. The following behaviors were then scored during every session: the time spent in each zone and the number of transitions between each zone. The median number and median duration of each period of object sniffing in each zone (sniffing an object was defined as follows: directing the nose to the object at a distance ≤2 cm and/or touching it with the nose) also were rated. In addition, the cumulative median time spent sniffing each object and the total median time spent sniffing all objects was calculated. The median number and duration of rearing and grooming in each zone were recorded.

Neuropathologic studies

Long-term pathologic changes were assessed in all the li-saline (n = 11) and li-pilo (n =11) rats subjected to behavioral testing. Brains were removed 5 days after the last behavioral testing. They were rapidly removed and frozen in isopentane chilled to –25°C and cut into 20-μm coronal sections. Quantification of cell density was performed with a 10 × 10 box, 1-cm2 microscopic grid on coronal sections stained with cresyl violet. The grid of counting was placed on a well-defined area of the cerebral structure of interest, and counting was carried out with a microscopic enlargement of 200- or 400-fold, defined for each single cerebral structure. Cell counts were performed bilaterally in three adjacent sections for each region by a single observer unaware of the animal treatment. The number of cells obtained in the six counted fields in each cerebral structure was averaged. This procedure was used to minimize the potential errors that could result from double counting, leading to overestimation of cell numbers. Neurons touching the inferior and right edges of the grid were not counted. Counts involved only neurons with cell bodies larger than 10 μm. Cells with small cell bodies were considered glial cells and were not counted. This method has been described elsewhere (32).

As we performed in all our previous studies concerning neuronal damage induced by li-pilo SE (32,33), brain sections were selected at three different levels [i.e., (a) the lateral thalamus and amygdala, (b) dorsal hippocampus and piriform cortex, and (c) ventral hippocampus and entorhinal cortex]. The anteroposterior level of the sections was selected according to stereotaxic coordinates of the rat brain atlas of Paxinos and Watson (34) and were –3.30 mm from bregma for the mediodorsal and lateral thalamus and medial and basolateral amygdala, –4.30 mm from bregma for the dorsal hippocampus (CA1, CA3 subfields, and the hilus of the dentate gyrus and piriform cortex), and –5.20 mm from bregma for ventral hippocampus and entorhinal cortex).

Statistical analysis

Most behavioral data did not follow gaussian distribution, and variances were not equal. Therefore we used medians and inferior and superior quartiles as statistical variables of central tendency and dispersion and nonparametric statistical analysis that is more powerful than parametric statistics in that case (35). Data from the li-pilo group were compared with those from the li-saline group by using a Mann–Whitney U test; a Friedman test for within-groups analyses, followed by a post hoc modified Mann–Whitney U test also was applied when adapted. Neuronal damage was expressed as the percentage of remaining neurons compared with control levels ± SEM. The significance of the differences observed between the li-pilo and li-saline groups was evaluated by using an analysis of variance followed by a Scheffé's t test for multiple comparisons.

RESULTS

Characteristics of lithium-pilocarpine–induced status epilepticus in adult rats

Control animals receiving lithium and saline did not exhibit any behavioral alteration. Rats subjected to lithium and pilocarpine developed the characteristic features of SE, which started ∼50 min after pilocarpine injection, as previously described (33,36,37). Among the 41 rats subjected to li-pilo, 38 (93%) rats developed SE; 18 rats died within the first 48 h after SE (13 between 2 and 6 h after onset of SE, five during the following night), and two rats died a few days after SE. Thus 21 (55%) of 38 rats survived SE. Within the 38 rats that developed SE, the mean latency to the first 4/5 stage seizure was 26.3 ± 6.8 min (mean ± SD), the latency to the onset of SE reached a mean value of 54.7 ± 8.1 min, and the mean duration between the first 4/5 stage seizure and SE was 27.7 ± 10.1 min. Among the 21 rats that survived li-pilo SE, 16 developed SRS, whose severity matched stages 4/5 of kindling, with a mean latency of 25 ± 7 days.

Behavioral studies

Elevated-plus maze

Three li-pilo–treated rats jumped from the apparatus after 20, 57, and 81 s, respectively, and were not included in the final analysis, which comprised eight li-pilo and 11 li-saline rats.

The total number of arm entries was significantly higher in the li-pilo group than in the li-saline group (U = 10.5, p = 0.004; Fig. 1A). This high locomotor activity was rather oriented toward open arms because the number of closed-arms entries was not different between the two groups (U = 32.5, p = 0.351; Fig. 1A), whereas the number of entries was significantly increased for open arms in the li-pilo group compared with the li-saline animals (U = 4.5, p < 0.001; Fig. 1A). The time spent on open and closed arms was significantly higher and lower, respectively, in li-pilo compared with li-saline rats (U = 8, p = 0.002 and U = 2, p < 0.001 for open and closed arms, respectively; Fig. 1B). The number of rearings was higher in the li-saline group (U = 9, p = 0.003; Table 1), but their median duration was similar in the two groups (U = 24.5, p = 0.109; Table 1). The li-pilo–treated rats made far more head-dips than li-saline rats (U = 1.5, p < 0.001; Table 1). The li-pilo group did not show any grooming behavior during the test, conversely to li-saline rats, which led to a highly significant difference between the two groups (U test, number and duration, U = 9.5, p = 0.003 and U = 7, p = 0.001, respectively; Table 1).

Figure 1.

Effect of li-pilo SE induced in adult rats on open and closed arm entries (A) and time spent on arms (B) in the elevated plus maze. Values represent medians, first and third quartiles, first and ninth deciles (bars), and extreme values (points) of each measure of 11 control rats and eight epileptic rats. **p < 0.01; ***p < 0.001; statistically significant differences between li-saline and li-pilo rats.

Table 1. Effect of li-pilo SE induced in adult rats on specific behaviors in the elevated plus-maze

Variable
Li-saline
group (n = 11)
Li-pilo
group (n = 8)
  1. Values represent median number or median duration (s), with quartiles in parentheses.

  2. ap < 0.01, bp < 0.001: statistically significant differences between li-pilo and li-saline rats.

Number of rearings18 (14.5–20.5)7 (5–9.25)a    
Number of head-dips8 (3–9)    24 (20.75–25.75)b
Number of groomings 2 (1.5–3.5) 0 (0–0.25)b   
Rearing duration1.6 (1.32–2.15)1.1 (0.82–1.68)   
Grooming duration  4 (3.17–6.14)0 (0–0.58)b   

The eight-arm maze

In the li-saline group, the total time necessary to enter all eight arms of the maze decreased over the 5 days of testing from 192 to 73 s (χ2, F(4, 11)= 25.81, p < 0.001; Fig. 2A). As expected, this group learned the task since the number of errors and the number of arms visited per session decreased (χ2, F(4, 11)= 18.50, p = 0.001; Fig. 2B and C), and the number of arms visited per min increased over the 5 days (χ2F(4, 11)= 24.51, p < 0.001; Fig. 2D). Conversely, the li-pilo group took approximately the same time to complete the task during the five sessions (171–230 s; χ2, F(4,11)= 2.50, p = 0.644; Fig. 2A). The total number of errors per session, the total number of arms visited per session, and the total number of arms visited per minute did not significantly differ between sessions (χ2, F(4,11)= 3.00, p = 0.557, χ2, F(4,11)= 3.00, p = 0.557; and χ2, F(4,11)= 5.75, p = 0.219, respectively; Figs. 2B-D).

Figure 2.

Effects of li-pilo SE induced in adults rats on the behavior in the eight-arm maze. Data represent the medians of 11 control rats and 11 epileptic rats. *p < 0.05; **p < 0.01; ***p < 0.001; statistically significant differences between li-saline and li-pilo rats.

Except for the first day, the total time per session was significantly higher in li-pilo compared with li-saline rats (7.10−5 < p < 0.008; Fig. 2A). The total number of errors per session and the total number of arms visited per session were significantly higher in the li-pilo than in the li-saline group for each session (5.10−4 < p < 0.044; Figs. 2B and C). The median number of arms entered per minute was 5.8 for the li-pilo group and 3.9 for the li-saline group during the first session; this number was significantly higher in li-pilo than in li-saline rats during sessions 2, 4, and 5 (0.001 < p < 0.016; Fig. 2D).

The object-recognition task

The total time spent sniffing all three objects was constant during the four sessions in the two groups (χ2F(3, 11)= 0.59, p = 0.901 and χ2, F(3, 11)= 1.80, p = 0.615 in li-saline and li-pilo rats, respectively). It was lower for the central object compared with the lateral ones. The time spent sniffing all objects and the number of sniffing sequences of all objects (data not shown) were never significantly different between the two groups in any of the four sessions. When taking each object individually, no difference in the time spent sniffing the central object was observed between the two groups (Fig. 3B). The same was observed for the left object except for session 3 (Fig. 3A), during which li-pilo rats spent significantly more time than li-saline rats sniffing this left object (U = 29.5; p = 0.040). The total time spent sniffing the right object during the first three sessions ranged from 11.0 to 11.8 s for li-saline and from 10.3 to 15.4 s for li-pilo rats (Fig. 3C). On day 4, when the novel (right) object was introduced, the two groups exhibited a longer time sniffing this object (18.9 and 24.6 s, respectively, for li-saline and li-pilo rats). The time spent sniffing this new object showed a statistically significant variation over the four sessions in the li-pilo group (χ2, F(3, 11)= 9.44, p = 0.024 and χ2, F(3, 11)= 5.86, p = 0.119, respectively, for li-pilo and li-saline groups). In this group, session 4 was significantly different from sessions 1–3 (0.004 < p < 0.047). However, in the two groups, a comparable novelty preference could be observed, reflected by the median duration of each sniffing sequence for the right object, which was higher during session 4 compared with sessions 1–3 (0.001 < p < 0.023 and 0.001 < p < 0.016 for li-saline and li-pilo groups, respectively; data not shown). The total time spent sniffing the new object during session four (Fig. 3C) as the total number of sniffing behaviors (data not shown) during this session were not significantly different between the li-pilo and li-saline group (U = 49, p = 0.478 and U = 39, p = 0.171, respectively).

Figure 3.

Effects of li-pilo SE induced in adults rats on sniffing behaviors in an object-recognition task. Values represent the median time of sniffing the left (A), central (B), and right (C) object of 11 control rats and 11 epileptic rats. *p < 0.05; statistically significant differences between li-saline and li-pilo rats.

Concerning locomotor activity, the number of transitions between the three delimited zones of the cage (Fig. 4A) remained unchanged during the four sessions in both groups (χ2, F(3, 11)= 1.51, p = 0.679 for the li-pilo group and χ2, F(3, 11)= 5.68, p = 0.127 for the li-saline group). However, the level of transitions was lower in the li-saline than in the li-pilo group. This difference was statistically significant for all sessions (U = 12, p < 0.01 for the first session; U < 5, p < 0.001 for sessions 2–4). The total time spent rearing (Fig. 4B) and the total number of rearings (data not shown) were similar in both groups and remained constant within each group over the four sessions. Finally, grooming activity was totally absent in epileptic rats and 9–24 times lower than in li-saline rats. The difference between the two groups was statistically significant in each session (2.10−6 < p < 0.005, Fig. 4C).

Figure 4.

Effects of li-pilo SE induced in adults rats on three behavioral measurements in an object-recognition task. Values represent medians of the number of transitions between each zone of the maze (A), total time of rearing (B), and total time spent grooming (C) of 11 control rats and 11 epileptic rats. **p < 0.01; ***p < 0.001; statistically significant differences between li-saline and li-pilo rats.

Histologic observations and neuronal loss

In the present study performed on cresyl-violet stained sections taken from rats having experienced 5 months of epilepsy (Fig. 5), two structures showed intense or total neuronal loss either with no tissue replacement (hole) or with swollen tissue replacement consisting mainly of filaments and glial cells bodies; these were the piriform cortex (70% and 90% cell loss in layers II and III–IV, respectively) and the entorhinal cortex (∼50% loss in layers II and III–IV). The amygdala underwent a marked loss of neurons (70% in the basolateral nucleus, 47% in the medial nucleus). Neuronal loss occurred also in the hippocampus, 53% in the hilus, ∼30% in pyramidal cell layer CA3 and 67% in CA1. Neuronal dropout reached 58% in the lateral thalamus with more shrunken neurons (Fig. 5). Neuronal loss was statistically significant in every structure analyzed (at least p < 0.01).

Figure 5.

Number of neurons remaining in selected areas of adult rats that underwent li-pilo SE as adults and studied 5 months after SE (n = 11). Values, expressed as percentage of control levels (n =11), represent mean ± SEM. *p < 0.05; **p < 0.01; ***p < 0.001; statistically significant differences between the li-pilo and control group. HILUS, hilus of the dentate gyrus; CA1, CA3 pyramidal cell layers of the hippocampus; BLA AMY, basolateral amygdala; Me AMY, medial amygdala; LAT AMY: lateral amygdala; LAT THAL: lateral thalamus; MD THAL: mediodorsal thalamus; PIR II, layer II of the piriform cortex; PIR III–IV, layers III–IV of the piriform cortex; ENT II, layer II of the entorhinal cortex; ENT III–IV, layers III–IV of the entorhinal cortex.

DISCUSSION

The data of the present study represent the first report of the behavioral consequences of SRSs lasting for a period as long as 5 months. In accordance with previous studies, spatial learning was severely impaired: animals were unable to learn the eight-arm maze. Avoidance for open arms in the elevated-plus maze was abolished. Nonetheless, li-pilo–treated rats performed well during the object-recognition task. The most damaged areas were the hippocampal CA1 area, the hilus, piriform and entorhinal cortices, and the amygdala.

Spatial memory

As reported in previous studies performed a few weeks after the occurrence of SRSs induced by pilo or li-pilo SE (17,22,24), the present data confirm that rats subjected to li-pilo SE and experiencing a 5-month duration of epilepsy are unable to learn the task in the radial-arm maze. The same types of results were reported in rats rendered epileptic by KA-induced SE (21,23,38). In the present study, no improvement of the performance of epileptic rats was noted along the 5 days of testing, neither for total time nor for total number of errors per session. The absence of learning and goal-oriented behavior also was reflected by a higher number of arms visited per session in the li-pilo than in the li-saline group. This difference is indicative of lower performance and poor strategy, and directly in line with the longer time needed by li-pilo compared with li-saline rats to achieve the task. In this maze, li-pilo rats did not appear hyperactive. These data indicate that spatial working memory was adversely affected, as previously reported (17).

The hippocampus has been reported to be prominently involved in spatial working memory (18,26,27,39). CA1 and CA3 pyramidal neurons as well as intact connections between these two regions appear critical in learning and retrieval of spatial memory (40,41). Indeed, neurons from layers II and III of the entorhinal cortex provide the entorhinal input to the hippocampus (42), and numerous studies report a prominent role of the entorhinal cortex in memory (for review, see ref. 43). The combination of lesions of the hippocampus and entorhinal cortex, observed in li-pilo rats, causes impairment of learning and retention (44), which may support the poor performance of these rats in the radial-arm maze in the present study.

Compared with previous neuropathologic data from our laboratory collected after a 2-month period of SRSs (33), a 3 months longer duration of epilepsy significantly worsens neuronal loss in the CA1 subfield of the hippocampus and layer II of the entorhinal cortex. Hence, it appears possible that not only SE but also SRSs, which both differently contribute to neuronal loss, might contribute to behavioral impairments in the li-pilo model of TLE. In the Morris Water Maze task, impaired learning appears as early as 10 days after KA-induced SE, which suggests that the long-term deficits of KA seizures are due to SE rather than to SRSs (45). Conversely, in the li-pilo model, the latency to reach the platform in the Morris Water Maze increases with the occurrence of SRSs (20). Thus impaired spatial learning may be the result of both SE-induced damage and SRSs. It seems that in both humans (46) and animals (47), the deterioration of cognitive functions could relate to the duration of the epilepsy. In humans, the association between the severity of structural damage and the duration of epilepsy, and lifetime number of seizures, was reported in both hippocampus (48) and entorhinal cortex (49). Further studies testing the performance of rats along the time course of the chronic period would be necessary to clarify this issue.

Anxiety

The present data represent the first report on the consequences of li-pilo–induced SE and epilepsy on the performance of rats in the elevated-plus maze. In a recent study, Kubova et al. (50) reported that all rats that they subjected to li-pilo SE as adults jumped off the maze and did not perform the task. In the present study, only three of 11 rats jumped off the maze, whereas the other eight completed the task. This difference could be linked to the time at which performance was assessed, 3 months after SE in the study by Kubova et al. (50) and 5 months in the present study. In the elevated-plus maze, li-pilo–treated rats showed increased overall activity reflected by a higher number of entries in open arms, whereas entries in closed arms remained unchanged. The open-arm activity is usually taken as the measure of the anxiety level (28). Moreover, the median number of head-dippings and rearings was higher in the li-pilo–treated group, which rather reflects disinhibited hyperactive behavior (51). Our results are in line with data obtained after a ventral lesion of the hippocampus, which potentially mediates the unconditioned fear response resulting from the exposure to a threatening situation, such as the elevated-plus maze (52). In the present model of TLE, lesions are as extensive in the ventral as in the dorsal hippocampus (53,54), and large lesions also are observed in numerous amygdala nuclei, mainly the basolateral, and in the entorhinal cortex (32,33,36,54, the present study). The ventral part of the hippocampus and the entorhinal cortex have bidirectional connections with the amygdala (55), which suggests that these structures may organize fear expression as a single integrated system (52). The disruption of these networks in li-pilo–treated rats might give rise to a misevaluation of threatening situations, which could in turn reduce anxiety and/or enhance impulsive inadapted behavior.

Object recognition

To our knowledge, no published data concern lesional animal models of TLE and the object-recognition task. In this task, li-pilo rats recognized the novel object as well as li-saline rats did, as reflected by similar total time spent sniffing the novel object. As in the elevated-plus maze, increased locomotor activity was observed in li-pilo rats, as shown by a higher number of transitions between zones. Grooming was almost abolished in li-pilo rats, as previously reported (50), but all other behavioral variables were identical in both groups. These data suggest a similar habituation in the two groups.

Our data are in accordance with previous articles that support spared object-discrimination capacities after hippocampal damage (29,30,56). Moreover, when taking into account the extent of the lesions in the entorhinal and piriform cortices, our study supports the view that simple object discrimination might be independent of those structures. In rodents, object discrimination has been shown to rely on visual and olfactory cues (57). In this task, even if objects were thoroughly cleaned after each trial, the new object was made up of a different matter (plastic instead of wood), which might provide an olfactory cue. The olfactory bulb presents reciprocal connections with the entorhinal (58) and the piriform cortex (59). In the li-pilo and pilo models of TLE, the whole olfactory system is severely damaged, at the level of the piriform cortex and the anterior olfactory nuclei (14,36), which might disrupt olfactory skills. Alternately, there could be visual compensation via the spared perirhinal cortex (60) that has been proposed to be critically involved in visual-recognition memory (61,62) and may be sufficient to solve this simple task.

Finally, our task design might not rely entirely on simple object discrimination and pure working memory. According to reported impaired spatial learning abilities in li-pilo treated rats, we chose to change the right object after the third session of habituation. This could lead to overlearning of environmental cues and contamination of the task by reference memory. To try to prevent this phenomenon, we chose a 24-h intertrial delay because it was previously shown that this duration made discrimination more difficult for control rats (63).

In conclusion, the present study shows behavioral deficits occurring after 5 months of epilepsy in the li-pilo model of TLE. The new finding of this study concerns the normal performance of epileptic rats in object discrimination, which suggests that such abilities do not entirely relate on hippocampal integrity. Neuronal loss was prominent in regions such as the hippocampus, entorhinal cortex, and amygdala, which is in accordance with severe learning impairment in the eight-arm maze and disinhibited behavior in the elevated-plus maze. In addition, rats were quite hyperactive, which reflects a lack of goal-oriented activity. The present data confirm that the li-pilo model of TLE represents a useful tool to explore behavioral disorders associated with the disease as well as potential pharmacologic treatments to prevent cognitive decline.

Acknowledgments

Acknowledgment:  This work was supported by the Institut National de la Santé et de la Recherche Médicale (U 398) and the Fondation pour la Recherche Médicale. We thank E. Koning and A. Ferrandon for friendly, professional, and helpful technical assistance.

Ancillary