Pharmacological characterization of social isolation-induced hyperactivity
- First Online:
- Cite this article as:
- Fabricius, K., Helboe, L., Fink-Jensen, A. et al. Psychopharmacology (2011) 215: 257. doi:10.1007/s00213-010-2128-9
- 189 Downloads
Social isolation (SI) of rats directly after weaning is a non-pharmacological, non-lesion animal model based on the neurodevelopmental hypothesis of schizophrenia. The model causes several neurobiological and behavioral alterations consistent with observations in schizophrenia.
In the present study, we evaluated if isolated rats display both a pre-pulse inhibition (PPI) deficit and hyperactivity. Furthermore, the sensitivity of SI hyperactivity to antipsychotic was evaluated.
Rats were socially isolated or group-housed for 12 weeks starting on postnatal day 25. In one batch of animals, the PPI and hyperactivity response were repeatedly compared. Furthermore, we investigated the robustness of the SI-induced hyperactivity by testing close to 50 batches of socially isolated or group-housed rats and tested the sensitivity of the assay to first- and second-generation antipsychotics, haloperidol, olanzapine, and risperidone, as well as the group II selective metabotrobic glutamate receptor agonist (LY404039).
Socially isolated rats showed a minor PPI deficit and a robust increase in hyperactivity compared with controls. Furthermore, SI-induced hyperactivity was selectively reversed by all antipsychotics, as well as the potential new antipsychotic, LY404039.
SI-induced hyperactivity was more pronounced and robust, as compared with SI-induced PPI deficits. Furthermore, SI-induced hyperactivity might be predictive for antipsychotic efficacy, as current treatment was effective in the model. Finally, using LY404039, a compound in development against schizophrenia, we have shown that the hyperactivity assay is sensitive to potential novel mechanisms of action. Thus, SI-induced hyperactivity might be a robust and novel in vivo screening assay of antipsychotic efficacy.
KeywordsSocial isolation Hyperactivity Pre-pulse inhibition Antipsychotic Metabotrobic glutamate receptor mGLUR Schizophrenia
Schizophrenia is a complex psychiatric disorder that manifests in early adulthood and affects approximately two thirds of the affected individuals throughout their lifetime (McGrath et al. 2008). The underlying pathophysiology remains, to a large extent, unknown, but it is likely that a combination of genetic and environmental factors contribute to the development of the disease. Alongside negative symptoms and cognitive deficits, positive symptoms are core symptoms of schizophrenia. Positive symptoms are classified as behaviors which are not seen in unaffected individuals. They include delusions and hallucinations (both auditory and visual; Pearlson 2000; Schultz et al. 2007), and in contrast to negative symptoms and cognitive deficits, positive symptoms are relatively well treated with current antipsychotics (Fenton et al. 2003; Green et al. 2002; Murphy et al. 2006).
When developing a new animal model of psychiatric illnesses, such as schizophrenia, it is inevitable to have an assay, e.g., a behavioral read-out that enables assessing symptoms of schizophrenia in animals. However, in order to validate this, or any other assay for the purpose of translating findings from the animals to the patient situation, the assay should be demonstrating predictive, and construct and face validity. In respect to predictive validity, a number of in vivo assays have indeed been shown to be predictive for antipsychotic efficacy such as the conditioned avoidance response (CAR) assay (Arnt 1982; Wadenberg and Hicks 1999) or amphetamine-induced hyperactivity (AIH; Schaefer and Michael 1984). However, these assays lack construct validity, as they utilize “healthy” rats or an acute pharmacological response and thus do not link up to our current understanding of the pathophysiological mechanisms of schizophrenia, such as the developmental aspect of the disease.
Rearing rats in social isolation (SI) from weaning until adulthood is a non-pharmacological non-lesion manipulation that originates on the developmental hypothesis of schizophrenia and would therefore increase the construct validity of a suitable behavioral assay used to predict antipsychotic efficacy in these animals. Indeed, rearing rats in social isolation induces pronounced effects on behavior (for review see Fone and Porkess 2008) and brain development (Fabricius et al. 2010b). In short, it has been reported that socially isolated rats display neurochemical alterations such as a hyper-responsive dopamine system, including increased responsiveness to psychomotor stimulants (Douglas et al. 2003; Fabricius et al. 2010a; Hall et al. 1998; Jones et al. 1992), and behavioral alterations such as disrupted sensory–motor gaiting shown as deficits in pre-pulse inhibition (PPI) (Cilia et al. 2001; Cilia et al. 2005; Geyer et al. 1993; Powell et al. 2003; Swerdlow et al. 2000), impaired memory and learning (Abdul-Monim et al. 2003; Bianchi et al. 2006; Douglas et al. 2003), as well as a consistent spontaneous hyperlocomotion to novelty (Weiss et al. 2000). PPI deficits, especially, in SI animals have been extensively investigated for predictive validity. Thus, it has been shown that SI-induced PPI deficits can be reversed with both typical and atypical antipsychotic treatment but not with antidepressant drugs such as diazepam or amitryptyline (Bakshi et al. 1998; Cilia et al. 2001; Domeney and Feldon 1998; Geyer et al. 1993; Nakato et al. 1997; Varty and Higgins 1995). Despite the extensive pharmacological characterization of the SI-induced PPI deficits, the robustness of these deficits has been questioned (Domeney and Feldon 1998; Weiss et al. 1999; Weiss et al. 2000). Indeed, a number of different factors, such as rat strain-related differences (Weiss et al. 2000), housing conditions (e.g., solid cage floors compared with grid floors) (Weiss et al. 1999), handling (Krebs-Thomson et al. 2001), first para-isolation period (Bakshi and Geyer 1999) and repeated testing (Domeney and Feldon 1998; Geyer et al. 1993; Varty and Higgins 1995) have been found to affect or even abolish PPI deficit in this model.
In contrast, the hyperactive locomotor response to a novel environment has been described as the most consistent finding in SI animals (for review see Fone and Porkess 2008), but only very limited pharmacological characterization has been performed (Shigemi et al. 2010).
In the present report, we present data from 3 years in 45 cohorts which were reared in groups of five (GH, group-housed) or in social isolation from weaning (postnatal day 25).
We expand these studies with an investigation of the robustness of the locomotor hyperactivity to repeated testing.
We also report the effects of first- and second-generation antipsychotics, as well as a selective metabotrobic glutamate 2/3 receptor (mGluR2/3) agonist LY404039, a drug currently in development against schizophrenia, in the SI hyperactivity assay.
All experiments were carried out in accordance with the European Communities Council Directive (86/609/EEC) for the care and use of laboratory animals and the Danish legislation regulating animal experiments. The Danish Animal Experimentation Inspectorate approved the protocols (journal no. 2004/561-798 and 2009/561-1596).
Treatment of animals was done according to procedures previously described (Cilia et al. 2001). Briefly, only male Lister-hooded rats (Harlan, The Netherlands) arrived at the in-house facilities on postnatal days (PND) 8–9 with foster mothers. To reduce the influence of differences in maternal care, all animals were cross-fostered randomly at birth, and litter sizes were standardized to 11 pups each. Upon arrival, dams and pups were left undisturbed until PND 25, when the pups were randomly assigned to be housed singly (SI) or in groups of five rats (GH controls). Housing consisted of a transparent, high, grid-lid macrolon cage, type III or IV, with wood chip bedding (type IV GH; 54 × 36 × 18 cm; type III SI; 37 × 22 × 18 cm). No enrichment was provided in either housing groups. All animals were housed under controlled temperature (22 ± 1.5°C) and humidity conditions (55–65%) with a 12:12 h light/dark cycle (lights on at 6:00 am). Handling and noise were kept to a minimum, and cages were cleaned twice a week for GH and once a week for SI. No further handling procedures were administrated during the first 8 weeks of isolation. Food (Altromin 1,324 pills, Brogaarden, DK) and water were available ad libitum. SI and GH animals were kept in the same room facility and could smell, hear, and see each other, but could not get in close contact to each other. All behavioral testing was conducted during the light conditions between 9:00 am and 5:00 pm after a minimum of 8 weeks of isolation.
Locomotor activity was tested after 8 weeks isolation in separate animal cohorts each consisting of a minimum of 18 SI and 15 GH. In total, 45 cohorts were tested in the locomotor activity assay.
To evaluate the consistency of the SI-induced hyperactivity, one cohort of animals was tested three times after 10, 12, and 14 weeks of isolation, respectively. In between 12 and 14 weeks, the animals were tested in the PPI assay (see PPI section for details).
For acclimation, all rats were brought to the laboratory 1 day prior to the test. On the test day, animals were placed in individual test cages and left undisturbed for 2 h (120 min). Each test cage (macrolon type III) was placed in a U-frame equipped with two rows of four infrared light sources and photocells. To avoid counts induced by stationary movements of the rat, recording of motility and rearing count required the interruption of adjacent light beams in the lower or upper row, respectively. Between each run, all cages were changed, so that each rat was placed into a fresh test cage. Those group-housed were marked on tails only once, directly after the first activity testing. Similar tail-handling was done to SI animals. Registration and timing of locomotor and rearing activity was fully automated (custom-designed hardware and software by Ellegaard Systems A/S, Faaborg Denmark).
The pharmacological experiments were conducted on randomly chosen animal cohorts after a minimum of 12 weeks and maximum of 16 weeks isolation. Animals in these studies were randomly assigned to either vehicle or drug in both housing groups (SI or GH n = 3–12 per group). Haloperidol (0.04 and 0.01 mg/kg), risperidone (0.16 and 0.08 mg/kg), olanzapine (0.16 and 0.08 mg/kg), and LY404039 (1.75 and 2.5 mg/kg) or vehicle were injected subcutaneously 30 or 60 min (LY404039) before commencing in the previously described locomotor activity setup.
One day after the locomotor experiment (see above) and 12 weeks after isolation, 14 GH and 14 SI rats were tested for two consecutive days in a PPI regime. PPI testing was performed using the Startle Monitor System (Kinder Scientific, USA), which consisted of eight sound-attenuated startle chambers and design based on computer software. During testing, animals were placed in an adjustable holder positioned 1 cm above the sensing platform. The PPI regime consisted of a 5-min acclimation period with white background noise (70 dB), followed by eight startle-pulse (120 db, 40 ms duration) presentations and 12 blocks of each of the following six trials in a randomized sequence: no pulse, startle-pulse, pre-pulse (4, 8, or 12 db above background noise and 30 ms duration) + startle-pulse, or finally pre-pulse (12 db above background noise) alone. Following these 12 blocks, another eight startle pulses were presented for the startle habituation calculations. The inter-stimulus interval was set at 100 ms, whereas the inter-trial interval varied between 9 and 15 s. In order to insure that the pre-pulses did not induce a startle by themselves, 11 pre-pulse alone trials were included in the trial protocol but were not used in the PPI calculations. To estimate the habituation to the startle, the mean response to the last eight startle pulses was expressed as the percent change of the mean response to the first eight startle pulses. The startle magnitude was calculated as the average response to all startle-pulse trials.
PPI was calculated as percent PPI for each pre-pulse intensity using the following equation: [100* (startle-pulse/pre-pulse + startle-pulse/startle-pulse)]. A low percentage score indicates a deficit in PPI.
Haloperidol (Sigma, UK), risperidone (Janssen Research US), olanzapine (H. Lundbeck A/S, DK), and LY 404039 (Eli Lilly, UK) were first dissolved in a minimum volume of HCL or methansulfoacid prior to final dilution in 0.9% NaCl (haloperidol, olanzapine, and risperidone), or isotonic (glucose) water (LY404039) and administrated subcutaneously at a volume of 1 ml/kg. All doses are expressed as base equivalent.
To test for differences in locomotor activity, each SI and GH cohort was compared using a Student's t test (data not shown). In addition, statistical differences over all cohorts were evaluated utilizing a two-way analysis of variance (ANOVA) with cohorts and housing (SI or GH) as independent variables. To investigate the effect of repeated testing, a two-way repeated measure (RM) ANOVA was used, with group (SI and GH cohort) and time (days 1 to 3) as factors. Post hoc analysis was performed by Fisher's–LSD design.
In the pharmacology studies, a two-way ANOVA, using housing (GH or SI) and treatment (dose) as between-subject factors, was used to investigate statistical differences between mean locomotor activity of SI and GH rats and efficacy of any antipsychotic treatment. For post hoc analysis, a Fisher's–LSD design was applied to segregate possible statistical differences.
To test the effect of housing conditions on startle amplitude, habituation, and mean PPI, a one-way ANOVA for each parameter was used. A two-way RM ANOVA was run to investigate the effect of housing and different pre-pulse intensities (pre-pulse stimulus intensity × housing condition). For post hoc analysis, Holm–Sidak design was applied.
To correlate pre-pulse intensity with locomotors behavior, a Pearson product moment correlation was applied.
All data are expressed as mean±SEM. In all studies, a P value ≤ 0.05 was considered significant. All statistical analyses were performed with SigmaStat (ver 3.0. Systat®, USA).
In the repeated testing experiment, one cohort was tested three times within a short time period, and SI animals were significantly more active than GH controls throughout the testing period. This was shown by a significant main effect of housing [F(1,43) = 56.46, P < 0.001] and a significant interaction between time and housing [F(2,43) = 3.75, P = 0.032] between SI and GH animals. There was also a significant main effect on time [F(2,43) = 38.79, P < 0.001], which seems to be due to an overall reduction of activity over time, possibly due to habituation to the test environment (Fig. 1c). However, the rate of decline in activity with repeated testing was equivalent irrespective of housing condition, indicating that activity levels decline with age and weight gain.
Overall, all antipsychotics and LY404039 selectively reduced SI-induced hyperactivity.
Haloperidol treatment reversed the isolation-induced hyperactivity [F(2,59) = 29.64, P < 0.001], with SI animals generally more active than GH controls [F(1,59) = 50.49, P < 0.001]. However, there was no interaction between housing and treatment [F(2,59) = 2.39, P = 0.10]. Post hoc analysis revealed that haloperidol was attenuating activity of SI animals already at doses lower than the ones affecting GH animals. Thus, 0.04 mg/kg reduced activity of both groups (P < 0.001 for both groups), whereas 0.01 mg/kg only reduced SI-induced hyperactivity (P = 0.023).
However, risperidone could selectively reverse the SI-induced hyperactivity, as shown by a main effect of treatment [F(2,59) = 22.43, P < 0.001] and a significant housing × treatment interaction [F(5,59) = 3.42, P = 0.03]. In line with the haloperidol experiment, SI animals were, overall, more active than GH [F(1,59) = 69.87, P < 0.001]. Post hoc analysis revealed that 0.16 mg/kg attenuated activity in both SI (P < 0.001) and GH (P = 0.005) whereas the 0.08 mg/kg selectively reduced activity in SI rats (P < 0.001).
Olanzapine treatment reversed the SI-induced hyperactivity [F(2,31) = 14.43, P < 0.001]. Again, SI animals were, overall, more active than GH [F(1,31) = 25.07, P < 0.001], but with no significant interaction of housing and treatment [F(2,31) = 1.61, P = 0.22]. Similar to haloperidol treatment, a post hoc analysis revealed that olanzapine at the higher concentration doses (0.16 mg/kg) affected both SI (P < 0.001) and GH (P = 0.019) whereas the lower doses at 0.08 mg/kg selectively affected the SI animals (P < 0.001).
Analyzing the effects of the pre-pulse intensities and housing on PPI by means of a two-way RM ANOVA confirmed significant differences between SI and GH controls [F(1,28) = 4.93, P = 0.035] and revealed that there was a main effect of pre-pulse intensity [F(2,56) = 110, P < 0.001]. And, although post hoc analysis supported that the PPI difference was mainly driven by the highest pre-pulse intensity, as a significant deficit in SI males compared with GH controls was only found for this pre-pulse intensity [PP82, P = 0.016] but not for the others [PP74, P = 0.082; PP78, P = 0.15]; no statistically significant interaction between housing and pre-pulse intensity was observed [F(2,89) = 0.73, P = 0.49]. Thus, the different PPI levels of housing seem not to depend on the level of pre-pulse intensity.
To investigate if there was a correlation between percent PPI deficits and activity after 12 weeks isolation, a Pearson correlation was applied using the data from the animals used in the PPI experiment shown above. No correlation was found between percent average PPI and total locomotor activity [P = 0.12, r = −0.30] or between the highest pre-pulse intensity (PP 82) and activity [P = 0.16, r = −0.27] (data not shown). Furthermore, no correlation was found between the individual housing groups in either the highest PP intensity or average PPI [GH average PPI, P = 0.63; SI average PPI, P = 0.57; GH PP82; P = 0.73, SI PP82; P = 0.96] (data not shown). Although no correlation was observed in the current data set, the sample size (SI, n = 16; GH, n = 14) is not regarded as large enough to exclude that a positive correlation might be obtained with a larger sample size.
Using a standardized isolation setup of Lister-hooded male rats, we reported consistent locomotor hyperactivity data from 45 cohorts of SI and GH male rats.
We expanded these studies showing that the SI-induced hyperactivity is robust to repeated testing and consequently repeated and frequent handling. In addition, we have shown the predictive validity of the hyperactivity assay by showing sensitivity to first- and second-generation antipsychotics. Finally, the potentially new antipsychotic LY404039, a selective group II metabotrobic glutamate receptor agonist currently in development for schizophrenia, was also effective in selectively reversing SI-induced hyperactivity at lower concentration doses.
PPI deficit in socially isolated rats and mice has been reported repeatedly across several laboratories for review, see (Weiss and Feldon 2001), and we were able to replicate these findings at least in one cohort that we report on in the present manuscript. However, we have found that the PPI deficits were not very marked or robust (see below) in contrast to the SI-induced hyperactivity. Indeed, these phenomena seem to have different underlying mechanisms, as we found no correlation between SI-induced hyperactivity and PPI in the cohort displaying both phenotypes. This is in agreement to what has been reported by other groups (Cilia et al. 2001; Cilia et al. 2005; Geyer et al. 1993; Varty and Higgins 1995; Weiss et al. 2000; Wilkinson et al. 1994).
Even though we found PPI deficits in the SI males, these deficits were not very marked and only significantly different in one of the tested pre-pulse intensities. Furthermore, we tested several other cohorts in our PPI regime but were not able to consistently show a PPI deficit (data not shown), with less than 50% of the tested cohorts showing SI-induced PPI deficit. Nevertheless, one group reported robust SI-induced PPI deficits across 28 cohorts with a success rate of 86% (Cilia et al. 2005). The same group also reported that repeated handling or prior locomotor activity did not influence the SI-induced PPI deficits (Cilia et al. 2001). Consequently, SI-induced PPI deficits studies have been proposed as a non-pharmacological disease model with impaired sensory motor gaiting and pharmacological validation with both first- and second-generation antipsychotics suggesting that the model possesses predictive validity for antipsychotic efficacy (Cilia et al. 2001; Cilia et al. 2005). However, others found that SI-induced PPI is very sensitive to factors such as handling procedures during the isolation period (Krebs-Thomson et al. 2001) or testing of locomotor activity prior to PPI testing (Domeney and Feldon 1998). And thus, it is reasonable to speculate that locomotor activity testing 1 day prior to PPI testing, as done in the present study might have had an effect on the effect size of the SI-induced PPI deficit in our study. In line with this, others report that at least two PPI sessions are needed to find a SI-induced PPI deficit (Bakshi et al. 1998; Domeney and Feldon 1998). Thus, replication and robustness of SI-induced PPI deficits seems at least difficult and might be affected by prior handling and locomotor testing, which questions if the SI-induced PPI assay would be suitable as a screening tool for antipsychotic efficacy. In contrast, SI-induced hyperactivity was very robust in our laboratory. Previous handling or repeated testing did not attenuate SI-induced hyperactivity, indicating that it might be better suited for screening purposes.
In line, hyperactivity is one of the most repeated observations in the isolated rats (for review see Fone and Porkess 2008). It is surprising that, to the best of our knowledge, no systematic pharmacological validation of the SI-induced hyperactivity has been published. As a consequence, we have tested the first-generation antipsychotic haloperidol as well as the second-generation antipsychotics olanzapine and risperidone and found that they could selectively reverse the isolation-induced hyperactivity, providing predictive validity to the SI-induced hyperactivity assay.
The antipsychotic effect of the neuroleptics, such as haloperidol, risperidone, and olanzapine have mainly been linked to their inhibitory effect on the dopamine D2 receptor (Kapur et al. 1999; Kapur et al. 2000) in striatal areas of the brain. Indeed, the effect of these neuroleptics in classical assays used to predict antipsychotic efficacy, such as CAR or AIH, is also correlated to their D2 receptor affinity (Arnt 1995; Wadenberg et al. 1990; Wadenberg and Hicks 1999), although inhibition of the 5-HT2A receptor might also contribute to the effects of risperidone and olanzapine (Arnt and Skarsfeldt 1998; Wadenberg et al. 1998). Interestingly, many of the behavioral consequences of social isolation rearing and especially the SI-induced hyperactivity have been related to an increased striatal dopamine release (Robbins et al. 1996). Henceforth, several studies have reported that pharmacological and behavioral stimulation of the dopaminergic system by amphetamine/cocaine or stress produces higher increases in dopamine in the striatum of SI rats (Fulford and Marsden 1998; Hall et al. 1998; Heidbreder et al. 2000; Jones et al. 1992), although this does not seem to be paralleled by changes in striatal D2-receptor function and binding in SI rats (Del et al. 2004). Nevertheless, the data outlined highlight that changes of the SI hyperactivity stem from changes in the dopamine signaling in striatum, and it is plausible to assume that the observed effect of the neuroleptics, haloperidol, risperidone, and olanzapine on SI-induced hyperactivity are driven by their D2 receptor blockade. This finding validates the assay to be predictive for antipsychotic efficacy with D2-receptor mechanism.
To evaluate if the social isolation model was also sensitive to novel and potentially antipsychotic mechanisms such as glutamatergic targets, we evaluated the selective metabotrobic glutamate 2/3 receptor agonist LY404039. A previous study with this drug suggested that it modulates glutamatergic activity in limbic and forebrain areas relevant for psychiatric disorders such as anxiety and psychosis (Rorick-Kehn et al. 2007). Furthermore, a study by Fell et al. (2008) found antipsychotic-like effects of LY404039 since it was able to block phencyclidine-induced hyperlocomotion in wild-type mice but not in mGluR2-deficient mice. These results demonstrate that LY404039 is mechanistically distinct from first- or second-generation antipsychotic drugs and is dependent on functional mGlu2 and not mGlu3 receptors (Fell et al. 2008). However, several studies using mGluR 2/3 agonists do not show marked effects on non-pharmacologically induced locomotion (Cartmell et al. 1999; Helton et al. 1998; Schlumberger et al. 2009).
The pharmacological results from the SI rats are valuable in the understanding of the neuronal pathways implicated in the SI-induced hyperactive behavior. As these results indicate, it is not only the dopamine system that seems involved but also the glutamatergic system. A recent study of isolation-induced hyperactivity revealed that the mGluR2/3 agonist LY379268 could reverse these isolation-induced impairments, thus supporting the findings in the present study (Jones et al. 2010). These results not only validate the social isolation model with respect to predictive validity but also the hyperactivity assay as an efficient procedure to elucidate distinct mechanisms of possible new targets for antipsychotic drugs.
In conclusion, we found that socially isolated Lister-hooded males display a modest PPI deficit and a robust SI-induced hyperactivity. The predictive validity of the latter was demonstrated by its sensitivity to antipsychotic treatment. Furthermore, using LY404039, a compound in development against schizophrenia, we have shown that the assay is sensitive to potential novel mechanisms of action, e.g., mGluR2/3 agonism. Hence, we suggest that SI-induced hyperactivity as a non-pharmacological animal screening model for predicting antipsychotic efficacy. It seems to be favorable in comparison to SI-induced PPI deficits in terms of robustness and possesses a higher level of construct validity as compared with standard screening assays, such as AIH or CAR, as it seems more relevant to utilize a “diseased” animal model which has a theoretical starting point in the developmental hypothesis for schizophrenia. However, evaluation of such a heterogeneous, multi-faceted disorder as schizophrenia in a single animal model is limited. Nonetheless, drug-induced changes in locomotor activity may have translational relevance to some of the positive symptoms of schizophrenia, but does not assess either the negative or cognitive deficits found in schizophrenia, for which there are currently no effective treatments. Thus, more studies are needed to fully investigate the potential of the social isolation model, but it is unlikely that a single animal model will be able to cover the whole aspect of such a heterogeneous disorder.
We sincerely thank technician Ditte Bekker-Jensen and senior technician Vibeke Nielsen H. Lundbeck A/S, Department of In Vivo Neuropharmacology, for performing the locomotor activity studies on the repeated tested cohort and performing the PPI study.
The work presented is part of a PhD work funded by the Neurocluster scholarship at the Faculty of Health and Science, University of Copenhagen, Denmark.