Introduction

Autism spectrum disorders (ASD) are characterized by qualitative abnormalities in reciprocal social interaction, communication, and repetitive and stereotyped patterns of behavior. Epilepsy is seen in about 30% of individuals with ASD. Conversely, in epilepsy populations there is an ASD prevalence of about 32%. In spite of this strong correlation, there is so far little experimental evidence for any direct causal relationship between seizures and ASD (Clarke et al. 2005; Spence and Schneider 2009).

Tuberous Sclerosis Complex (TSC) is caused by heterozygous mutation in either the TSC1 or TSC2 gene. About 25% of individuals with TSC meet criteria for classic infantile autism, about 50% for ASD, and 70–90% have a lifetime history of epilepsy (Smalley 1998; Bolton et al. 2002). Approximately 30–40% of TSC patients have global intellectual disability (IQ < 70) (Joinson et al. 2003). When the total TSC population is studied, epilepsy shows strong correlations with intellectual disability and ASD (Smalley 1998; Gomez et al. 1999; de Vries et al. 2007).

Animal models reduce the complex scenarios of neuropsychiatric disorders to more simply defined variables and are as such useful to study mechanisms of pathology (Fisch 2007). To date, three TSC animal models, Tsc1 +/−, Tsc2 +/− knockout mice and spontaneous mutation Tsc2 +/− (Eker) rats (Eker and Mossige 1961), have been analyzed for the neurobiological basis of behavioral phenotypes. None of the models exhibit spontaneous seizure activity. Although there is substantial variation between the different animals, all models express changes in learning and memory with the Tsc1 +/− mice showing deficits in social behavior as well (Waltereit et al. 2006; Goorden et al. 2007; Ehninger et al. 2008). Taking together the current human and animal data, results suggest that seizures may not be necessary to cause ASD (de Vries and Howe 2007; Goorden et al. 2007), and that there might be direct genetic effects (Smalley 1998; de Vries and Howe 2007). However, social deficits have only been reported in one TSC animal model (Goorden et al. 2007), and it is not known whether seizures may be sufficient to cause social deficit behaviors in TSC or animal models (Spence and Schneider 2009).

To examine the effect of TSC genotype and epilepsy on learning and memory and social behaviors, we studied Tsc2 +/− (Eker) and wild-type rats. To examine the impact of seizures during development, we treated wild-type and Tsc2 +/− rats with kainic acid (KA) injections at postnatal day 7 (P7) and P14. All animals responded with a typical crescendo-like status epilepticus that lasted many hours (Sayin et al. 2004). Animals of all four groups (naïve wild-type; naïve Tsc2 +/−; epilepsy wild-type; epilepsy Tsc2 +/−) were analyzed for behavioral changes at the age of 3–6 months (Fig. 1).

Fig. 1
figure 1

Experimental design. Tsc2 +/− and wild-type rats were challenged with status epilepticus by KA injections at P7 and P14. Naïve rats had not undergone the epilepsy paradigm. Behavior was analyzed at the age of 3–6 months

Full size image

Experimental procedures

Animals

Rats were housed under a 12 h/12 h day–night cycle. Only male rats were used for experiments. Tsc2 +/− (Eker) and wild-type genotypes were determined by PCR (Rennebeck et al. 1998). In all experiments, littermates with similar distribution to Tsc2 +/− (Eker) and wild-type were used. All testing took place during the day phase. All experimental procedures were performed according to permission from local state authorities (Regierungspräsidium Karlsruhe).

KA-induced status epilepticus

KA monohydrate (Sigma, Deisenhofen, Germany) was dissolved in phosphate-buffered saline. At P7, male offspring received an intraperitoneal injection with 3 mg/kg KA, and at P14, a second injection with 4 mg/kg KA. The animals were returned to the home cage with their mother. About 15 min after injection, a crescendo-like status epilepticus started in all injected animals, and lasted for several hours. For this study, always a Tsc2 +/− parent was crossed with a wild-type parent. During the first status epilepticus, there was 17.5% mortality, during the second one, there was no mortality. Of the animals surviving the epilepsy paradigm, 39% revealed the Tsc2 +/− (Eker) mutation after genotyping; of all naïve rats, 44% were Tsc2 +/−. Late-onset seizures were observed neither during animal care nor behavioral analysis.

Open field

The arena had an area of 52 × 52 cm2, a height of 45 cm and was made of grey polyvinylchloride (PVC). Light intensity was 50 lux. Movements were recorded with a digital video camera and analyzed on a personal computer using Biobserve Viewer (Biobserve, Bonn, Germany) tracking software. Rearings were detected by light barriers at a height of 11 cm. At the beginning of the 30 min session, animals were placed in the center of the arena. Data were analyzed as 5 min bins.

Light/dark-box

The apparatus had an area of 75 × 25 cm2, a height of 40 cm and was made of grey PVC. The dark compartment had an area of 25 × 25 cm2, was separated from the light compartment by a grey PVC wall with a 10 cm wide, 15 cm high alleyway and was covered by a grey PVC plate. The light compartment was illuminated with 100 lux. Rats were initially placed in the dark, closed compartment and allowed to habituate for 1 min to the apparatus. The alleyway was then opened, movements in the light compartment were recorded with a digital video camera for 10 min and analyzed manually by a trained observer.

Novel object recognition

Behavior was recorded with a digital video camera and analyzed manually by a trained and blinded observer (Schneider et al. 2008). Objects were made of metal or glass and existed in duplicate. Rats were habituated to the open field for 10 min, 24 h before behavioral testing. The test consisted of an initial 3-min sample phase (P1) and a 3-min discrimination phase (P2) that were separated either by an intertrial interval of 15 min or by an interval of 24 h. During P1, the rat was placed in the centre of the open field and exposed to an unknown object (A). After cessation of P1 the rat was returned to the homecage and the object was removed. The rat was placed back in the open field after 15 min or 24 h for object discrimination during P2 and was then exposed to the familiar object (A′, an identical copy of the object presented in P1) and a novel test object (B). Exploration of the objects (sniffing, touching and gnawing) was recorded during P1 and P2. Sitting beside or standing on top of the objects was not scored as object investigation. Testing for the 15 min interval was done 1 day before the 24 h interval test and different objects were used for both tests.

Fear conditioning and extinction

The apparatus and tracking system was as described earlier (Waltereit et al. 2008), with the exception of using a chamber designed for rats as context A, area 25 × 30 cm2, height 33 cm (H10-11R-TC, Coulbourn Instruments, Allentown, GA, USA). Context B was a standard home cage without litter. The conditioned stimulus (CS) was a 5000 Hz sinus wave of 30 s duration. The unconditioned stimulus (UCS) was a scrambled footshock of 0.5 mA and 1 s duration, which was applied during the last second of the CS. Day 1: Habituation. 10 min exploration in context A. Day 2: Conditioning in context A. 3 min exploration, 30 s CS-UCS, 3 min exploration, 30 s CS-UCS, 3 min exploration. Day 3: Recall of conditioning. 6 min exploration in context A, 1 h interval, 3 min exploration in context B, 3 min exploration in context B with CS. Day 4: Extinction. 3 min exploration in context A, followed by 15 times 30 s CS and 3 min exploration in context A. Day 5: Recall of extinction. Recall of conditioning. 6 min exploration in context A, 1 h interval, 3 min exploration in context B, 3 min exploration in context B with CS.

Morris water maze

The pool had a diameter of 145 cm, a height of 50 cm and was made from black PVC. Rats were trained in the water maze with extra-maze cues (water made opaque by addition of 3 L milk, water temperature 26 ± 1°C, 10 × 10 cm2 platform with white rubber surface, height of the platform 13 cm with its top surface 2 cm below water level, maximum trial duration 120, 15 s on platform at the end of trials). Illumination was 30 lux. Animals were trained for 5 days three times a day with 2 h intervals (Waltereit et al. 2006). After a total of 15 training trials, the probe trial was performed with the platform removed (probe trial duration = 60 s). During training trials, animals were assessed for the latency to escape to the hidden platform. During the probe trial, animals were assessed for the time spent in the four quadrants and for the frequency with which they crossed the position of the platform in the target quadrant (and the respective position in the three non-target quadrants). Swim paths were recorded using the EthoVision video tracking system (Noldus Information Technology, Wageningen, The Netherlands).

Social interaction

The test was performed in the open field (Schneider et al. 2008). Social partners were 6 week old male Long-Evans rats (Charles River, Sulzfeld, Germany). All animals were habituated for 5 min to the arena 24 h before testing. The experimental animal was first placed into the test arena and was allowed to habituate for 1 min before the social partner was introduced. The following behavioral elements were videotaped and quantified by a trained and blinded observer only for the experimental rats: (A) Social behavior: contact behavior, social exploration and approach/following were scored as social behaviors. (1) Contact behavior: contact behavior includes (a) grooming (chewing and licking the partner’s fur) and (b) crawling over/under the partner; (2) social exploration: (a) anogenital investigation (sniffing or licking the anogenital area of the social partner) and (b) non-anogenital investigation (sniffing at any part of the partner’s body, except the anogenital area); (3) approach/following: approaching or following the social partner in the test arena. (B) Evade: running, leaping or swerving away from the social partner. Evade, which is normally defined as a defensive behavior in the context of social play, was scored in the social interaction test as an active withdrawal from social contact.

Statistics

Analyzes (two-way analysis of variance (ANOVA) or two-way repeated measures ANOVA, respectively, followed by Bonferroni post-hoc tests) were performed with SigmaStat software (Systat Software, San Jose, CA, USA). Differences were considered statistically significant if P < 0.05. Graphical artwork was created with Prism (GraphPad Software, San Diego, CA; USA) and CorelDraw software (Corel Corporation, Ottawa, Canada). Graphs always show mean, standard error of the mean (SEM) and significant results from the two-way ANOVA or two-way repeated measures ANOVA, respectively. Significant results from Bonferroni post-hoc tests are indicated in the graphs as well. One asterisk in a graph represents P < 0.05, two asterisks P < 0.01, three asterisks P < 0.001.

Results

We first evaluated general behavioral parameters. Locomotor activity was assessed in the open field. There were no differences in distance travelled over time (Fig. 2a, two-way repeated measures ANOVA for experimental groups [F (3,230) = 0.9826, P > 0.05], time [F (5,230) = 190.2, P < 0.001] and interaction [F (15,230) = 0.3975, P > 0.05]), but Tsc2 +/− rats showed less rearings than wild-type animals (Fig. 2b, two-way ANOVA for genotype [F (1,46) = 4.320, P < 0.05], epilepsy [F (1,46) = 1,293, P > 0.05] and interaction [F (1,46) = 0.07021, P > 0.05]). Anxiety was analyzed using the light/dark-box. The latency to enter the light compartment was longer in rats which had undergone the epilepsy paradigm (Fig. 2c, two-way ANOVA for genotype [F (1,52) = 0.5466, P > 0.05], epilepsy [F (1,52) = 4.941, P < 0.05] and interaction [F (1,52) = 0.4554, P > 0.05]). Next, we examined novel object recognition and exploration. All animals spent a higher proportion of time exploring the novel objects during the recall phase, indicating that they recognized the previously presented object, and there were no differences between experimental groups after an interval of 15 min (Fig. 3a, two-way ANOVA for genotype [F (1,52) = 0.04816, P > 0.05], epilepsy [F (1,52) = 0.1031, P > 0.05] and interaction [F (1,52) = 0.02086, P > 0.05]) and an interval of 24 h (Fig. 3b, two-way ANOVA for genotype [F (1,54) = 0.5534, P > 0.05], epilepsy [F (1,54) = 0.001447, P > 0.05] and interaction [F (1,54) = 0.08213, P > 0.05]). However, Tsc2 +/− rats spent less time than wild-type animals exploring novel objects during the acquisition phase (Fig. 3c, two-way ANOVA for genotype [F (1,54) = 5.732, P < 0.05], epilepsy [F (1,54) = 0.01435, P > 0.05] and interaction [F (1,54) = 0.5991, P > 0.05]).

Fig. 2
figure 2

Tsc2+/− reduces exploratory behavior and developmental epilepsy increases anxiety. a Distance travelled in the open field. Wild-type naïve n = 16, Tsc2+/− naïve n = 13, wild-type epilepsy n = 12, Tsc2+/− epilepsy n = 9. b Rearings in the open field. Wild-type naïve n = 16, Tsc2+/− naïve n = 13, wild-type epilepsy n = 12, Tsc2+/− epilepsy n = 9. c Latency to light in the light/dark-box. Wild-type naïve n = 16, Tsc2+/− naïve n = 13, wild-type epilepsy n = 15, Tsc2+/− epilepsy n = 12. Data are expressed as mean and (b, c) SEM. * P < 0.05

Full size image
Fig. 3
figure 3

Novel object recognition. a Novel object preference (15 min interval). Wild-type naïve n = 15, Tsc2 +/− (Eker) naïve n = 12, wild-type epilepsy n = 17, Tsc2 +/− (Eker) epilepsy n = 12. b Novel object preference (24 h interval). Wild-type naïve n = 16, Tsc2 +/− (Eker) naïve n = 13, wild-type epilepsy n = 17, Tsc2 +/− (Eker) epilepsy n = 12. c Object exploration time (mean values from animals used both 15 min and 24 h interval experiments). Wild-type naïve n = 16, Tsc2 +/− naïve n = 13, wild-type epilepsy n = 17, Tsc2 +/− epilepsy n = 12. Data are expressed as mean and SEM. * P < 0.05

Full size image

We then started a series of experiments to assess learning and memory. Fear conditioning is a form of classical conditioning. Our protocol tested both contextual and auditory cue conditioning, and also analyzed extinction of these memories. After conditioning with two CS-UCS presentations, all rats demonstrated learning of both contextual (Fig. 4a, two-way ANOVA for genotype [F (1,51) = 0.04820, P > 0.05], epilepsy [F (1,51) = 0.08456, P > 0.05] and interaction [F (1,51) = 0.9020, P > 0.05]) and auditory cue associations (Fig. 4c, two-way ANOVA for experimental group [F (3,100) = 0.1168, P > 0.05], auditory cue [F(1,100) = 84.68, P < 0.001] and interaction [F (3,100) = 0.1482, P > 0.05]). There were no differences between experimental groups. After extinction of the associations by presenting the CS without UCS for 1 h, rats showed some reduction in both contextual (Fig. 4b, two-way ANOVA for genotype [F (1,51) = 2.308, P > 0.05], epilepsy [F (1,51) = 0.3967, P > 0.05] and interaction [F (1,51) = 0.0000205, P > 0.05]) and auditory cue conditionings (Fig. 4d, two-way ANOVA for experimental group [F (3,100) = 0.6257, P > 0.05], auditory cue [F (1,100) = 47.11, P < 0.001] and interaction [F (3,100) = 0.02162, P > 0.05]), but there were no differences between groups. The Morris water maze tests spatial memory, a form of hippocampus-dependent learning. There were no differences in latency to escape to the platform during training trials (Fig. 5a, two-way repeated measures ANOVA for experimental groups [F (3,700) = 1.390, P > 0.05], trials [F (14,700) = 52.61, P < 0.001] and interaction [F (42,700) = 1.103, P > 0.05]). During a probe trial after the last training trial, we analyzed swimming in the target quadrant (Fig. 5b, two-way ANOVA for experimental group [F (3,100) = 0.1293, P > 0.05], target [F (1,100) = 22.07, P < 0.001] and interaction [F (3,100) = 0.6895, P > 0.05]) and swimming over the target platform position (Fig. 5c, two-way ANOVA for experimental group [F (3,54) = 0.2630, P > 0.05], target [F (1,54) = 34.55, P < 0.001] and interaction [F (3,54) = 0.3329, P > 0.05]). All animals had learned to search in the target region, and there were no differences between groups. Swim speeds during the probe trial were without differences between groups (mean ± SEM): Wild-type naïve 20.58 cm/s ± 1.33, Tsc2 +/− naïve 20.19 cm/s ± 1.44, wild-type epilepsy 20.92 cm/s ± 1.16, Tsc2 +/− epilepsy 20.25 cm/s ± 0.43. Wild-type naïve n = 8, Tsc2 +/− naïve n = 7, wild-type epilepsy n = 8, Tsc2 +/− epilepsy n = 8. Two-way ANOVA for genotype [F (1,27) = 0.6446, P > 0.05], epilepsy [F (1,27) = 0.8616, P > 0.05] and interaction [F (1,27) = 0.9019, P > 0.05].

Fig. 4
figure 4

Fear conditioning and extinction are not altered. a Freezing after contextual conditioning. Wild-type naïve n = 16, Tsc2 +/− naïve n = 13, wild-type epilepsy n = 15, Tsc2 +/− epilepsy n = 11. b Freezing after extinction of contextual conditioning. Wild-type naïve n = 16, Tsc2 +/− naïve n = 13, wild-type epilepsy n = 15, Tsc2 +/− epilepsy n = 11. c Freezing after auditory cue conditioning. Wild-type naïve n = 16, Tsc2 +/− naïve n = 13, wild-type epilepsy n = 15, Tsc2 +/− epilepsy n = 11. d Freezing after extinction of auditory cue conditioning. Wild-type naïve n = 16, Tsc2 +/− naïve n = 13, wild-type epilepsy n = 15, Tsc2 +/− epilepsy n = 11. Data are expressed as mean and SEM

Full size image
Fig. 5
figure 5

No learning and memory deficit in the Morris water maze. a Latency to platform during training trials. Wild-type naïve n = 15, Tsc2 +/− naïve n = 13, wild-type epilepsy n = 14, Tsc2 +/− epilepsy n = 12. b Time in quadrants during probe trial. Wild-type naïve n = 15, Tsc2 +/− naïve n = 13, wild-type epilepsy n = 14, Tsc2 +/− epilepsy n = 12. c Platform crossings during probe trial. Wild-type naïve n = 8, Tsc2 +/− naïve n = 7, wild-type epilepsy n = 8, Tsc2 +/− epilepsy n = 8. Data are expressed as mean and SEM

Full size image

Finally, we performed a series of experiments to assess various social behaviors in the open field with a young adolescent wild-type rat as social partner. Firstly, social exploration was examined using anogenital and non-anogenital exploration and approach & follow behaviors. Summary scores of these three social exploration tasks showed that naïve Tsc2 +/− rats had reduced social behaviors (Fig. 6a) and that epilepsy induced reduction in social exploration behaviors in the wild-type rats to rates similar to the naïve Tsc2 +/− rats (Fig. 6a, two-way ANOVA for genotype [F (1,27) = 3.201, P > 0.05], epilepsy [F (1,27) = 6.691, P < 0.05] and interaction [F (1,27) = 1.923, P > 0.05], significant Bonferroni post-hoc tests: epilepsy within wild-type P < 0.01, genotype within naïve P < 0.05). These differences were predominantly attributable to reduction in non-anogenital exploration (Fig. 6b, two-way ANOVA for genotype [F (1,27) = 0.9666, P > 0.05], epilepsy [F (1,27) = 0.9666, P > 0.05] and interaction [F (1,27) = 1.223, P > 0.05]) rather than to significant change in anogenital (Fig. 6c, two-way ANOVA for genotype [F (1,27) = 2.839, P > 0.05], epilepsy [F (1,27) = 16.35, P < 0.001] and interaction [F (1,27) = 1.527, P > 0.05], significant Bonferroni post-hoc tests: epilepsy within wild-type P < 0.001) or approach and follow behaviors (Fig. 6d, two-way ANOVA for genotype [F (1,27) = 3.073, P > 0.05], epilepsy [F (1,27) = 1.073, P > 0.05] and interaction [F (1,27) = 1.115, P > 0.05]). Next we examined social evade and contact behavior. Social evade is the active avoidance of contact with the social partner. Naïve rats in both groups showed similarly low rates of evade. Rats which had undergone the epilepsy paradigm showed increased social evade (Fig. 6e, two-way ANOVA for genotype [F (1,27) = 2.467, P > 0.05], epilepsy [F (1,27) = 15.97, P < 0.001] and interaction [F (1,27) = 1.471, P > 0.05], significant Bonferroni post-hoc tests: epilepsy within Tsc2 +/− P < 0.01). Contact behavior is a further type of normal social interaction in rodents. Epilepsy also reduced contact behavior in wild-type and Tsc2 +/− rats (Fig. 6f, two-way ANOVA for genotype [F (1,27) = 0.008507, P > 0.05], epilepsy [F (1,27) = 9.642, P < 0.01] and interaction [F (1,27) = 0.1144, P > 0.05], significant Bonferroni post-hoc tests: epilepsy within wild-type P < 0.05).

Fig. 6
figure 6

Tsc2+/− and developmental epilepsy lead to deficits in social behavior. a Total social exploration (total of b–d). Wild-type naïve n = 8, Tsc2+/− naïve n = 7, wild-type epilepsy n = 8, Tsc2+/− epilepsy n = 8. b Anogenital exploration. Wild-type naïve n = 8, Tsc2+/− (Eker) naïve n = 7, wild-type epilepsy n = 8, Tsc2+/− (Eker) epilepsy n = 8. c Non-anogenital exploration. Wild-type naïve n = 8, Tsc2+/− (Eker) naïve n = 7, wild-type epilepsy n = 8, Tsc2+/− (Eker) epilepsy n = 8. d Approach and follow. Wild-type naïve n = 8, Tsc2+/− (Eker) naïve n = 7, wild-type epilepsy n = 8, Tsc2+/− (Eker) epilepsy n = 8. e Social evade. Wild-type naïve n = 8, Tsc2+/− naïve n = 7, wild-type epilepsy n = 8, Tsc2+/− epilepsy n = 8. f Contact behavior. Wild-type naïve n = 8, Tsc2+/− naïve n = 7, wild-type epilepsy n = 8, Tsc2+/− epilepsy n = 8. Data are expressed as mean and SEM. * P < 0.05, ** P < 0.01, *** P < 0.001

Full size image

Discussion

Our study revealed that neither the Tsc2 +/− (Eker) mutation, KA-induced status epilepticus at P7 and P14 nor combination of these paradigms induced learning and memory deficits in rats. However, both the Tsc2 +/− mutation and the epilepsy paradigm independently caused autistic-like social deficit behaviors with reduced novel object, environmental and social exploration behaviors in the naïve Tsc2 +/− rat, and increased anxiety, social evade, reduced social exploration and contact behavior in Tsc2 +/− and wild-type rats after experimental epilepsy.

The epilepsy paradigm induced status epilepticus on two separate occasions for several hours during development. Late-onset seizures were observed neither during animal care nor behavioral analysis. We did not assess the animals by electroencephalography (EEG) or electrocorticography (ECoG) to detect subtle or non-convulsive epileptic discharges during adult age. However, because the epilepsy paradigm did not induce any learning and memory deficits, it seems unlikely that long-lasting or late-onset epileptic activity confounded the other results.

The findings for the naïve Tsc2 +/− rats are in agreement with our previous study, which found no learning and memory deficits in conditioned taste aversion, radial maze and the Morris water maze (Waltereit et al. 2006). Surprisingly, the experimental epilepsy paradigm did not induce learning and memory deficits. This is in contrast to a similar study which made also use of a status epilepticus paradigm (Sayin et al. 2004). The latter study used Sprague–Dawley rats (Dr. Stafstrom, personal communication), whereas our rats were bred on Long-Evans background. Thus, strain differences may explain the discrepant findings for wild-type rats. Learning and memory deficits in rats are not an equivalent of global intellectual disability in TSC patients. Nevertheless, it is surprising that the combination of Tsc2 haploinsufficiency and developmental seizures did not result in any learning and memory deficits. Global intellectual deficit in TSC (de Vries and Prather 2007) is strongly associated with prolonged seizures, infantile spasms (Joinson et al. 2003; O’Callaghan et al. 2004) and onset of seizures in the first year of life (Jozwiak et al. 1998; Gomez et al. 1999; Bolton et al. 2002; Curatolo et al. 2008). An explanation for the discrepancy between these TSC patients and our model could lie in the duration, type or developmental timepoint of the experimental seizures. We therefore acknowledge the possibility that the epilepsy paradigm used here may not have been of sufficient duration or intensity to induce global intellectual impairment, that KA-induced status epilepticus may not model the pathology of infantile spasms adequately, and that the time point of epilepsy induction could also have been too late in development given, that P7 and P14 in rats is later in brain development than seizures during the first 6–12 months of human life. These would be important experimental parameters to modify in future studies.

Our experiments focused on social interaction and anxiety-related behaviors rather than on communication or repetitive and stereotyped patterns of behavior, the other two diagnostic domains of ASD (Tordjman et al. 2007; Viding and Blakemore 2007). Tsc2 +/− rats displayed reduced exploratory behaviors. In the open field, Tsc2 +/− rats exhibited less rearings, which can be interpreted as reduced exploration of the environment, given that locomotor behavior was unchanged. Object exploration time in the novel object recognition task was reduced. Naïve Tsc2 +/− rats also showed less social exploration. In a similar pattern, Tsc1 +/− KO mice showed less interaction with a social partner and reduced nest building behavior, interpreted as analogous to autistic-like behavior in patients with TSC (Goorden et al. 2007). In contrast, Tsc2 +/− KO mice were not impaired in exploration and social behavior tests, which was explained by a modifier gene hypothesis (Ehninger et al. 2008). In the light/dark-box, wild-type and Tsc2 +/− rats which had undergone the epilepsy paradigm expressed increased anxiety. The study by Sayin and colleagues which made similar use of KA-induced status epilepticus during development, also described increased anxiety in adult rats in the elevated plus maze (Sayin et al. 2004). The social interaction test is a test traditionally used to study anxiety-related behavior in animals due to its sensitivity to both anxiolytic and anxiogenic effects (File 2000; File et al. 2001; Irvine et al. 2001) and is therefore thought to present a model of social anxiety in humans (File and Hyde 1978). In humans, social anxiety and ASD are often seen in conjunction (Moldin and Rubenstein 2006). Moreover, impaired reciprocal social interaction is a core deficit in ASD. These tests are therefore recognized as ASD phenotype tasks (Crawley 2007; Moy et al. 2007). The epilepsy paradigm reduced social exploration in adult rats. In addition, social evade and contact behavior tasks also showed significant impairments induced by the epilepsy paradigm. This reduction might thus be related to the increased anxiety-related response towards the unfamiliar social partner, but it could also indicate a general performance deficit in appropriate social behavior. Taken together, both Tsc2 haploinsufficiency and epilepsy independently lead to autistic-like social deficits. It is of note that the nature of social deficits was not identical in the Tsc2 +/− and developmental epilepsy groups. In the human ASD literature, there have been suggestions of subtle phenotypic differences between ASD with and without epilepsy. Children with ASD and epilepsy, for instance, showed significantly reduced social interaction with peers of a similar age (Turk et al. 2009). We suggest that in TSC the neurobiological abnormalities caused by gene mutation may be sufficient to lead to some autistic-like social deficit behaviors, and that seizures may have a direct and additive effect by inducing further social deficits to increase the likelihood and range of autistic-like behaviors.

The apparent dissociation between learning and memory and social deficits is of further interest. The results suggest the possibility of a differential threshold of vulnerability—that is, fewer seizures may be required to induce social deficits and that more, prolonged or developmentally earlier seizures may be required to lead to learning and memory deficits. Returning to ASD, our results suggest that epilepsy in general may induce social, but not necessarily learning and memory deficits in individuals who have ASD or who are at risk of ASD. This may help to explain the increased rates of ASD in epilepsy populations and the significantly increased rates of ASD in those with genetic syndromes associated with a high risk of ASD.