INTRODUCTION

The immediate-early gene early growth response 3 (Egr3) is associated with schizophrenia risk (Kim et al, 2010; Yamada et al, 2007; Zhang R et al, 2012) and expressed at reduced levels in the brains of patients with the mental illness (Mexal et al, 2005; Yamada et al, 2007). Animal studies also support a role for Egr3 in schizophrenia pathogenesis. We have previously reported that Egr3-deficient (Egr3−/−) mice display locomotor hyperactivity, a phenotype associated with schizophrenia (Gainetdinov et al, 2001), which is reversed by treatment with either haloperidol or clozapine (Gallitano-Mendel et al, 2008). However, the response of the mice to these two medications was distinctly different. Whereas the dose of haloperidol that normalized the hyperactivity of Egr3−/− mice did not affect the locomotion of wild-type (WT) control animals, the dosage of clozapine required to normalize the activity of Egr3−/− mice profoundly suppressed the locomotor activity in their WT littermates (Gallitano-Mendel et al, 2008). This relative resistance to the locomotor suppression effects of clozapine, compared to controls, parallels the heightened tolerance of schizophrenia patients to the side effects of antipsychotics (Cutler, 2001). The cause of this effect in either humans or Egr3−/− mice is not known. Like in humans, identification of the neurobiological cause of abnormal behaviors in gene-deficient mice can be remarkably challenging. Indeed, prior histological studies failed to identify differences in levels of neurotransmitter receptors in the brains of Egr3−/− mice to explain these abnormalities (Tourtellotte and Milbrandt, 1998). This points to a need for methods to identify receptor differences that underlie behavioral and pharmacological abnormalities in genetically altered mice.

Clozapine remains one of the leading antipsychotic medications to date, yet its mechanism of action remains unknown. This is due, in part, to its complex receptor binding profile. Clozapine binds to a wide range of receptors in the brain, including numerous subtypes of dopamine, serotonin, histamine, adrenergic, and muscarinic receptors (reviewed in Meltzer and Huang (2008), Roth et al (2004), and Stahl (2008)). The mechanism underlying the sedating effects of clozapine in humans is also unknown, although it is frequently attributed to antagonism of the histamine H1 receptor (Casey, 1997). We hypothesized that, by systematically testing a range of pharmacological compounds, which bind to subsets of receptors in the clozapine binding profile (ie, a ‘pharmacological dissection’), we could identify the receptor subtype responsible for the resistance of Egr3 −/− mice to the locomotor activity suppression by clozapine. This should shed light on the mechanism underlying the sedating effects of clozapine in humans. In addition, this method should identify a neurotransmitter receptor defect in Egr3−/− mice, providing a clue into a neurobiological abnormality of schizophrenia patients, while also revealing the next step of our hypothesized pathway of schizophrenia susceptibility genes (Gallitano-Mendel et al, 2008).

In this study, we show several novel findings resulting from this approach. First, the H1 histamine receptor is not responsible for the resistance of Egr3−/− mice to locomotor suppression by clozapine. Second, we demonstrate that the locomotor activity response of Egr3−/− mice appears to distinguish second-generation antipsychotics (SGAs, also known as ‘atypical antipsychotics’) from first-generation antipsychotics (FGAs, also known as ‘typical antipsychotic’). Third, we show that selective antagonists for the serotonin 2A receptor (5HT2AR) suppress the locomotor activity of WT, but not Egr3−/−, mice and thus mimic the effect of clozapine. Finally, we find that Egr3−/− mice have a nearly 70% decrease in 5HT2AR binding in the prefrontal cortex (PFC) and display a blunted behavioral head-twitch response to 5HT2AR agonist 1-(2,5-dimethoxy 4-iodophenyl)-2-amino propane (DOI). These findings suggest that action at the 5HT2AR contributes to the locomotor suppressive effects of clozapine and other SGAs in mice, and may play a role in the sedating effects of these medications in humans. Furthermore, the reduced levels of 5HT2ARs we identified in Egr3−/− mice parallel the results of numerous in vivo and post-mortem studies that report decreased levels of 5HT2ARs in the frontal cortex of schizophrenia patients, including first-break, untreated individuals (Dean and Hayes, 1996; Erritzoe et al, 2008; Garbett et al, 2008; Hurlemann et al, 2008; Lopez-Figueroa et al, 2004; Matsumoto et al, 2005; Ngan et al, 2000; Rasmussen et al, 2010; Serretti et al, 2007). Taken together, these findings suggest a possible mechanism through which human Egr3 may influence susceptibility to schizophrenia.

MATERIALS AND METHODS

Animals

Previously generated Egr3−/− mice (Tourtellotte and Milbrandt, 1998) were backcrossed to C57BL/6 mice for more than 20 generations. Animals were housed on a 14/10 h light/dark schedule with ad libitum access to food and water. A large breeding colony of Egr3+/− × Egr3+/− mice was maintained to produce study animals. Studies were conducted on adult male littermate progeny of these matings.

Behavioral Testing

Behavioral testing was performed during daytime hours under ambient light conditions. Progeny male +/+ and −/− animals were identified as ‘matched pairs’ at the time of genotyping and added to experimental cohorts at a minimum of 2 months of age. A total of 10 independent cohorts of animals were used throughout the course of all behavioral studies. To accommodate IACUC recommendations for reduction of animal numbers, mice were used in an average of three tests before being euthanized by CO2 asphyxiation. All animals were killed by the age of 12 months, if not before. Between tests, mice underwent a washout period of greater than, or equal to, five drug half-lives, and were re-randomized into treatment groups, to minimize possible confounds from repeated use. Furthermore, testing for each drug was replicated in a second, independent cohort of mice. Locomotor activity effects were robust and replicable, reducing the likelihood of possible confounds secondary to re-use of animals. Specific sample sizes per group, per experiment are included in the figure legends for each study.

Activity Monitoring

The effect of pharmacological agents on the locomotor activity of WT and Egr3−/− mice was measured using the SmartFrame system (Kinder Scientific, Poway, CA) (Table 1). Drug dosages were selected following literature review and subsequent dose–response testing in pilot groups of WT C57BL/6 mice to establish the dosage that suppressed locomotor activity in WT mice (Supplementary Figure S1). The suppressive dosage, vehicle control, and intermediate dosages were then tested in a large cohort of matched Egr3−/− and WT littermates. All studies were replicated in a second, separate cohort of animals. The second of the replicate studies is presented in the Results section.

Table 1 Effect of Drugs Targeting Receptors Bound by Clozapine on the Activity of WT and Egr3−/− Mice
Full size table

Activity was evaluated in transparent (47.6 × 25.4 × 20.6 cm3 high) polystyrene enclosures using a computerized photobeam system (MotorMonitor Kinder Scientific). Animals were placed in the enclosures 20 min after drug administration, and activity was monitored for 1 h. Locomotor activity was calculated using a number of movements (total photobeam breaks) as the dependent variable for total activity. The term ‘Reduced’ is used for statistically significant decreases in locomotor activity, compared with vehicle-treated animals of the same genotype, which remain >1000 movements per h. Decreases in locomotor behavior below 1000 movements per h are labeled ‘suppressed’.

Data are depicted as graphs of average total activity in response to drug dosage for each genotype. In addition, the same data are also graphed to show the percentage decrease in activity compared with vehicle-treated controls of the same genotype. For the latter graph, each animal was compared to the average of all vehicle-treated animals of the same genotype to generate individual ‘percent change in activity’ values. These values were then averaged for all mice in a treatment group (defined by genotype and drug dosage) to produce bar graphs. Error bars in all graphs denote standard error of the mean for each treatment group.

Video Recording

After completion of testing, a subset of WT and Egr3−/− mice were video recorded following administration of FGA chlorpromazine (10 mg/kg), SGA olanzapine (3 mg/kg), and 5HT2AR-specific antagonist MDL-11939 (10 mg/kg). Littermate animals were administered either drug or vehicle and allowed to acclimate for 30 min before the removal of the cage lid for brief recording with a hand-held video camcorder. Gentle shaking of the cage by the investigator was used to stimulate the activity of immobile mice.

Drowsiness, Motor impairment, and Stereotypic Behavior Assessment

The effect of haloperidol (3 mg/kg) and clozapine (7 mg/kg) on drowsiness, motor impairment, and stereotypic behavior was assessed in a cohort of WT and Egr3−/− mice. Behavior was scored for a period of 2 min at 30 and 60 min post-drug administration. Abnormal movements were scored according to a behavioral checklist for stereotypy and dyskinesia, adapted from McNamara et al. (2006) and Khan et al. (2004). The categories scored included: grooming episodes, head bobbing, and myoclonic twitches of the abdomen, head/facial, and limb regions. The total number of head bobs, and face, trunk, or limb twitches in the 2 min period were summed to produce a total stereotypy count. Grooming episodes were rare across all groups, and therefore not included. A second assessment was performed using a sedation and motor impairment rating scale adapted from Aitchison et al. (2000). Scores represent the average of two independent observers blind to both genotype and treatment. A detailed protocol and rating scales are included in Supplementary Materials.

DOI-Induced Head-Twitch Response

Head-twitch response to DOI (1 mg/kg) was assessed as described previously (Gonzalez-Maeso et al, 2003). Animals were placed in a transparent polystyrene cage 15 min following administration of drug or vehicle and video recorded for 30 min at close range by a camera suspended above the cage. Head twitches were independently scored by two observers blind to genotype and treatment, and scores were averaged for statistical analysis.

Radioligand Binding Assay

Radioligand binding with [3H]ketanserin was used to measure the level of expression of 5-HT2AR in the PFC of drug-naive Egr3−/− and WT littermate control mice. Animals were killed via CO2 asphyxiation, brains were immediately removed, and the PFC was dissected from a coronal slice spanning from Bregma: 1.95, Interaural: 5.78 and Bregma: 0.00, Interaural: 3.80, using the Coronal C57BL/6J Atlas from the Mouse Brain Library (Rosen et al, 2000). Collected tissue was snap-frozen on dry ice and stored at −80 °C until binding studies were performed. [3H]Ketanserin (DuPont-NEN, Boston, MA) binding (0.0625–10 nM; 10 concentrations) to 5HT2AR was measured at equilibrium in 500 μl aliquots (50 mM Tris-HCl; pH 7.4) of membrane preparations (10–57 μg protein per tube), which were incubated at 37 °C for 60 min as described previously (Gonzalez-Maeso et al, 2008). Nonspecific binding was determined in the presence of 10 μM methysergide (Tocris Bioscience, Ellisville, MO), and ranged from 27±2 to 61±4% of total binding in all groups. The study was performed in two independent groups of animals; n=8 animals per genotype for each study.

Drug Preparation and Administration

Chlorpromazine, clozapine, haloperidol, ketanserin, and DOI were obtained from Sigma Aldrich (St Louis, MO). Olanzapine, quetiapine, and ziprasidone were obtained through the NIMH Chemical Synthesis and Drug Supply Program (Bethesda, MD). MDL-11939 was obtained from Tocris Bioscience (Ellisville, MO). Chlorpromazine and DOI were dissolved in saline. Olanzapine, quetiapine, and haloperidol were dissolved in a small amount of glacial acetic acid and further diluted in sterile water. Clozapine and MDL-11939 were dissolved in HCl and diluted in sterile water. Ketanserin was diluted in DMSO and sterile water. Ziprasidone was diluted in 45% 2-hydroxypropyl-β-cyclodextrin. Concentrated aliquots of each drug were stored at −20 °C. Aliquots were thawed at 37°C and diluted to their final concentration in sterile saline on the day of testing. Solutions were buffered as necessary to achieve a final pH of 6.5–7.5. Ki determinations were generously provided by the National Institute of Mental Health's Psychoactive Drug Screening Program, Contract No. HHSN-271-2008-00025-C (NIMH PDSP; Bethesda, MD).

For each drug tested, vehicle was prepared in an identical manner without the addition of drug. Drug or vehicle was administered via intraperitoneal injection in a 10 ml/kg volume.

Data Analysis

Statistical analyses, including analysis of variance (ANOVA), Student's t-test, and standard error of the mean (SEM) were performed in SPSS (Chicago, IL) and Microsoft Excel. Locomotor activity and DOI-induced head-twitch behavior were evaluated using a two-way ANOVA. Behavioral assessment data were examined in SPSS using repeated-measures multivariate ANOVA (MANOVA) with treatment and genotype as a between-subjects factor and time as a repeated measure. Data are represented as means±SEM in all graphs.

RESULTS

We have previously reported that Egr3−/− mice are resistant to the locomotor inhibitory effects of the antipsychotic medication clozapine ((Gallitano-Mendel et al, 2008), see online video at http://www.nature.com/npp/journal/v33/n6/extref/1301505x3.mov). This response is not due to the baseline hyperactivity (a schizophrenia-like rodent phenotype) displayed by Egr3−/− mice as the animals do not show this response to the antipsychotic haloperidol. We have previously shown that haloperidol normalizes the hyperactivity of Egr3−/− mice to WT vehicle-treated levels at a dosage that has no effect on the locomotor activity of WT mice, and higher dosages of haloperidol reduce the activity of both WT and Egr3−/− mice to the same degree (Gallitano-Mendel et al, 2008). In contrast, clozapine profoundly suppresses the activity of WT mice at a dosage that reduces the hyperactivity of Egr3−/− mice only to WT vehicle-treated levels (Gallitano-Mendel et al, 2008). Thus, the locomotor inhibition produced by haloperidol is not the same as that resulting from clozapine, and Egr3−/− mice distinguish this difference.

In this study, we have employed a ‘pharmacological dissection’ approach, using increasingly selective drugs to target specific receptor subtypes in the clozapine binding profile, to identify the receptor abnormality that is responsible for the decreased sensitivity of Egr3−/− mice to locomotor suppression by clozapine, compared with haloperidol. To identify drugs that mimic the effects of clozapine, we first established the dosage of each test drug that suppressed locomotor activity (ie, decreased activity below 1000 movements per h) in a pilot group of WT mice (Supplementary Figure S1). We then tested whether that dosage also reduced the locomotor activity of Egr3−/− mice.

Agents Selective for Receptors Commonly Associated with Sedation Fail to Replicate the Effect of SGAs in Egr3−/− Mice

We began by targeting the receptor systems to which the sedating effects of clozapine and other SGAs are commonly attributed: the histamine H1 receptor and α-adrenergic receptors (Alves et al, 2010; Casey, 1997; Mengod et al, 1996; Parsons and Ganellin, 2006). Dose–response pilot experiments in WT C57Bl/6 mice (the background strain of Egr3−/− mice) revealed that the selective H1 receptor antagonist pyrilamine (also known as mepyramine) does not reduce locomotor activity, even at doses up to 50 mg/kg, the highest dose used in mice found in the literature (Figure 1 and Table 1) (Parsons and Ganellin, 2006; Shishido et al, 1991). A pilot study with diphenhydramine, another relatively selective H1 antagonist (Parsons and Ganellin, 2006), yielded similar results (Table 1). Promethazine, a less selective H1 receptor antagonist (Wishart et al, 2008; Wishart et al, 2006) and member of the phenothiazine family, suppressed activity at the highest administered dose (50 mg/kg), but did so equally in both WT control and Egr3−/− mice, failing to reproduce the response to clozapine (Table 1). Tests with terazosin, an α1-adrenergic receptor antagonist, and medetomidine, an α2-specific agonist used as a sedative in veterinary medicine (Alves et al, 2010), likewise showed similar levels of locomotor suppression in Egr3−/− and WT mice (Table 1). These findings indicate that neither H1 histamine receptors nor α-adrenergic receptors alone are responsible for the resistance of Egr3−/− mice to the locomotor inhibitory effects of clozapine.

Figure 1
figure 1

Histamine H1 antagonism is not responsible for resistance of Egr3-deficient (Egr3−/−) mice to locomotor suppression by clozapine. Locomotor activity was monitored for 60 min in wild-type (WT) mice following administration of highly specific H1 receptor antagonist pyrilamine. Pyrilamine was not sedating at 10 or 50 mg/kg, the highest dose reported in the literature (Roth, 2008) (n=3 per group).

PowerPoint slide

Full size image

Egr3−/− Mice Exhibit Resistance to Locomotor Suppression by SGAS, but not FGAS

As the receptors commonly implicated in sedation did not appear to be responsible for the resistance of Egr3−/− mice to this effect of clozapine, we returned to our earlier finding that Egr3−/− mice do not display resistance to the locomotor inhibitory effects of the FGA haloperidol (Gallitano-Mendel et al, 2008). In other words, haloperidol reduces the hyperactivity of Egr3−/− mice to normal levels at a dosage that does not affect the activity of WT mice, and higher doses of the medication inhibit activity in WT and Egr3−/− mice in an equivalent manner (Gallitano-Mendel et al, 2008). We hypothesized that this may be because haloperidol is a high potency antipsychotic that is also less sedating than other FGA medications. We therefore tested whether chlorpromazine (0, 5, and 10 mg/kg), a low-potency FGA that is highly sedating, would produce a similar behavioral effect on Egr3−/− mice. Figure 2a shows that chlorpromazine reduced the activity of Egr3−/− mice at the same dosage as WT mice, an effect similar to that of haloperidol, and markedly different than that of clozapine. A two-way ANOVA revealed a main effect of chlorpromazine (F(2,54)=21.1; p<0.001) and genotype (F(1,54)=45.5; p<0.001) on locomotor activity, and a treatment by genotype interaction (F(2,54)=7.1; p<0.05). Figure 2b shows that each dose of chlorpromazine reduced activity to a similar degree, on average, in both WT (77% with 5 mg/kg and 95% with 10 mg/kg) and Egr3−/− mice (74% with 5 mg/kg and 89% with 10 mg/kg) (differences were not significant by Bonferroni-corrected Student's t-test). On visual inspection both WT control and Egr3−/− mice appeared immobile following the highest dose (10 mg/kg) of chlorpromazine (Supplementary Video S2).

Figure 2
figure 2

First-generation antipsychotics (FGAs) reduce activity to a similar degree in Egr3-deficient (Egr3−/−) and wild-type (WT) mice. The locomotor activity of Egr3−/− and WT mice was monitored for 60 min following administration of chlorpromazine, a low-potency, highly sedating FGA. As previously reported with haloperidol (Gallitano-Mendel et al., 2008), Egr3−/− mice demonstrate a similar susceptibility to locomotor suppression by chlorpromazine as WT controls (see also Supplementary Video S2). (a) Vehicle-treated Egr3−/− mice are hyperactive in comparison to vehicle-treated WT mice. Chlorpromazine reduced locomotor activity in a dose-dependent manner to a similar degree in both Egr3−/− and WT mice (n=10 per group). (b) The average activity of vehicle-treated mice for each genotype was used to calculate the percent decrease from basal activity for each animal (see Methods). The average percent decrease in activity is presented for both Egr3−/− and WT mice treated with either 0, 5, or 10 mg/kg chlorpromazine.

PowerPoint slide

Full size image

We then repeated this assay with other SGAs to assess whether Egr3−/− mice would show the same locomotor response to other medications within the same classification as clozapine. Figure 3a shows that olanzapine, an SGA designed to mimic the receptor binding activity of clozapine, suppressed the locomotor activity of WT mice to nearly zero, while the activity of Egr3−/− mice was decreased only to vehicle-treated WT levels (also see Supplementary Video S3). This result was identical to that of clozapine (Gallitano-Mendel et al, 2008). A two-way ANOVA evaluating locomotor activity following administration of olanzapine (0, 1, 2, and 3 mg/kg) revealed a main effect of treatment (F(3,56)=17.7; p<0.001) and genotype (F(1,56)=155.7; p<0.001), and a treatment by genotype interaction (F(3,56)=4.6; p<0.01). Figure 3b shows that each dose of olanzapine reduced the activity more in WT mice than in Egr3−/− mice. Compared to vehicle-treated mice, a 3 mg/kg dose of olanzapine reduced activity by 98%, on average, in WT mice, but only 51% in Egr3−/− mice (p<0.001, Student's t-test).

Figure 3
figure 3

Egr3-deficient (Egr3−/−) mice are resistant to locomotor inhibition by second-generation antipsychotic agents (SGAs). The locomotor activity of Egr3−/− and WT mice was monitored for 60 min following administration of SGAs. (a) Olanzapine suppressed the activity of WT controls at a dosage of 1 mg/kg, while a dosage of 3 mg/kg reduced the activity of Egr3−/− mice to normal WT activity levels (n=8 per group) (see also Supplementary Video S3). (b) The average activity of vehicle-treated mice for each genotype was used to calculate the percent decrease from basal activity for each animal (see Materials and Methods). The average percent decrease is presented for both Egr3−/− and WT mice treated with 0, 1, 2, or 3 mg/kg olanzapine. (c) Quetiapine reduced the activity of Egr3−/− mice to vehicle-treated WT activity levels, while abolishing almost all locomotor activity in WT mice, at 20 mg/kg (n=7 per group). (d) The average percent decrease in activity from vehicle group is presented for both Egr3−/− and WT mice treated with 0, 10, or 20 mg/kg quetiapine. (e) Ziprasidone (2.5 mg/kg) suppressed the activity of WT mice, while 5 mg/kg was required to reduce the hyperactivity of Egr3−/− mice to normal WT levels (n=10 per group). (f) The average percent decrease in activity from vehicle group is presented for both Egr3−/− and WT mice treated with 0, 2.5, or 5 mg/kg ziprasidone. *Significant post hoc comparisons of simple main effects between Egr3−/− and WT mice at the dose leading to an extreme suppression in activity in WT controls (a, c, and e), or Student's t-test after Bonferroni correction for multiple comparisons (b, d, and f) (*p<0.05; **p<0.001).

PowerPoint slide

Full size image

Similar analyses following administration of two additional SGAs replicated these effects. Figures 3c–f show that Egr3−/− mice are similarly resistant to the locomotor suppressive effects of quetiapine and ziprasidone as to clozapine and olanzapine. The two-way ANOVA following administration of quetiapine (0, 10, and 20 mg/kg) revealed a main effect of treatment (F(2,36)=20.4; p<0.001) and genotype (F(1,36)=92.5; p<0.001), and a treatment by genotype interaction (F(2,36)=4.0; p<0.05) (Figure 3c). Figure 3d shows that 20 mg/kg of quetiapine reduced activity more in WT than Egr3−/− mice when compared to vehicle-treated mice of the respective genotype (p<0.001, Student's t-test). Ziprasidone treatment (0, 2.5, and 5 mg/kg) also revealed main effects of treatment (F(2,41)=18.1; p<0.001) and genotype (F(1,41)=62.6, p<0.001), and a treatment by genotype interaction (F(2,41)=4.5, p<0.05) (Figure 3e). Figure 3f shows that, like the other SGAs, in comparison to vehicle-treated mice, each dose of ziprasidone reduce the activity of WT mice to a greater degree than the activity of Egr3−/− mice (p<0.005 for 2.5 mg/kg dose, p<0.001for 5 mg/kg dose, Student's t-test). These findings suggest that Egr3−/− mice differ from WT mice in their response to the locomotor suppressive effect of SGAs, but not to that of FGAs.

The SGAs differ from FGAs in producing a significantly lower incidence of extra-pyramidal side effects (Pierre, 2005). This response in mice is identified by ‘stereotypic movements’ (or ‘stereotypy’). As the activity-monitoring test we employed does not differentiate sedation from other causes of immobility, we evaluated the behavioral response of Egr3−/− and WT mice to the SGA clozapine, and the FGA haloperidol, using rating scales for drowsiness, motor impairment, and stereotypic behaviors. Figures 4a and b show that WT mice are significantly more sensitive to both the drowsiness and motor impairment caused by clozapine than are Egr3−/− mice. Repeated-measures MANOVA revealed a main effect of both treatment and genotype and a treatment by genotype interaction on drowsiness (F(1,28)=21.5 (p< 0.001), F(1,28)=7.6 (p<0.05), F(1,28)=10.4 (p<0.005), respectively) and motor impairment (F(1,28)=33.8 (p<0.001), F(1,28)=6.7 (p<0.05), F(1,28)=6.7 (p<0.05)). In contrast, WT and Egr3−/− mice did not differ in the number of stereotypic movements they displayed following administration of clozapine (Figure 4c). Analysis of stereotypy data (Figure 4c) revealed a main effect of clozapine treatment (F(1,28)=15.5 (p<0.005)), but not genotype (p>0.05). Within-subjects analysis revealed a time by genotype interaction on stereotypy (F(1,28)=5.2; p<0.05), indicating that the two genotypes varied in the timing of their stereotypic movements across the test period, with WT mice displaying more stereotypy at 30 than 60 min, and Egr3−/− mice showing a more level number of stereotypic movements between the two time points. However, this timing effect is unlikely to account for drug-induced differences in locomotor-activity between Egr3−/− and WT mice as both time points are included in the 60 min activity monitoring session.

Figure 4
figure 4

Stereotypic behavior does not account for the differential response of Egr3-deficient (Egr3−/−) mice to first-generation antipsychotics (FGAs) vs second-generation antipsychotics (SGAs). Drowsiness, motor impairment, and stereotypy scores were assessed at 30 and 60 min following administration of the drug (clozapine, 7 mg/kg, a, b, and c; or haloperidol, 3 mg/kg, d, e, and f) or corresponding vehicle. Egr3−/− and wild-type (WT) mice responded differently to clozapine than to haloperidol in measures of drowsiness (a, d) and motor impairment (b, e), but not stereotypy (c, f) (n=8 per group for clozapine; n=9 per group for haloperidol). *Significant comparisons between vehicle and drug treatment groups within genotype (*p<0.05; **p<0.01; ***p<0.005; by Student's t-test).

PowerPoint slide

Full size image

Like clozapine, haloperidol also induced a different degree of drowsiness in Egr3−/− mice than in WT mice. However, the effect was the opposite to that of clozapine, with Egr3−/− mice showing more drowsiness than WT mice following haloperidol administration (Figure 4d). Haloperidol caused motor impairment in both Egr3−/− and WT mice, although the difference in the response of the two genotypes was not evident until 60 min after drug administration (Figure 4e). Repeated-measures MANOVA on haloperidol treatment revealed a main effect of both treatment and genotype on drowsiness (F(1,32)=33.8 (p<0.001) and F(1,32)=17.2 (p<0.001), respectively) and motor impairment (F(1,32)=116.3 (p<0.001) and F(1,32)=4.6 (p<0.05), respectively) and a treatment by genotype interaction on drowsiness (F(1,32)=15.5 (p<0.001)), but not on motor impairment (p>0.05). Analysis of stereotypy data (Figure 4f) revealed a main effect of treatment (F(1,32)=70.0 (p<0.001)), but not of genotype (F(1,32)=0.007 (p=0.9)). Within-subject analysis revealed a main effect of time on drowsiness (F(1,32)=5.3; p<0.05) and a three-way time by dose by genotype interaction (F(1,32)=5.33; p<0.05). These results indicate that Egr3−/− mice differ from WT mice in their sensitivity to sedating and motor-impairing effects of antipsychotic medications, but they do not differ in their sensitivity to the stereotypic effects of these drugs. This suggests that the different motor effects of FGAs vs SGAs on Egr3−/− mice are not stereotypic in nature.

5HT2AR Antagonists Parallel the Effect of Clozapine on Egr3−/− Mice

One of the leading features distinguishing SGAs from FGAs is the high affinity SGAs display for the 5HT2AR (Meltzer et al, 2003). We therefore tested whether this receptor was responsible for the resistance of Egr3−/− mice to the locomotor suppressive effects of these medications by examining the effect of drugs with relatively selective affinity for the 5HT2AR. First, we examined the effect of the 5HT2AR antagonist ketanserin. Figure 5a shows that 5 mg/kg ketanserin suppressed locomotor activity in WT mice, while the locomotor hyperactivity of Egr3−/− mice was not even reduced to vehicle-treated WT levels. A two-way ANOVA on activity following treatment with 5HT2AR antagonist ketanserin (0, 2.5, and 5 mg/kg) revealed a main effect of treatment (F(2, 42)=4.5; p<0.05) and genotype (F(1,42)=65.8; p<0.001), and no treatment by genotype interaction (F(2,42)=1.1; p>0.05). Figure 5b shows that ketanserin reduced activity more in WT mice than Egr3−/− mice. Compared to vehicle-treated mice, a 5 mg/kg dose of ketanserin reduced activity by 78%, on average, in WT mice, but only 17% in Egr3−/− mice (p<0.001, Student's t-test).

Figure 5
figure 5

Serotonin 2A receptor (5HT2AR) antagonists suppress the locomotor activity of wild-type (WT), but not Egr3-deficient (Egr3−/−) mice. Locomotor activity was monitored for 60 min following administration of 5HT2AR-specific agents or vehicle. (a) Ketanserin suppresses the locomotor activity of WT mice at 5 mg/kg, a dose that decreases the hyperactivity of Egr3−/− mice, but does not reduce it to normal WT levels (n=7–9 per group). (b) The average activity of vehicle-treated mice for each genotype was used to calculate the percent decrease from basal activity for each animal (see Methods). The average percent decrease in activity from vehicle group is presented for both Egr3−/− and WT mice treated with 0, 2.5, or 5.0 mg/kg ketanserin. (c) MDL-11939 (10 mg/kg) suppresses the activity in WT mice, but fails to reduce the hyperactivity of Egr3−/− mice to normal WT activity levels (n=8 per group) (see also Supplementary Video S4). (d) The average percent decrease in activity from vehicle group is presented for both Egr3−/− and WT mice treated with 0, 2.5, 5, and 10 mg/kg MDL-11939. *Significant post hoc comparisons of simple main effects between Egr3−/− and WT mice at the dose leading to an extreme reduction in activity in WT controls (a and c) or Student's t-test after Bonferroni correction for multiple comparisons (b and d) (*p<0.005; **p<0.001).

PowerPoint slide

Full size image

Ketanserin binds with high affinity to 5HT2ARs, but also binds to other serotonin receptors, as well as H1 and D1 dopamine receptors, with lower affinity, as summarized in Table 2. We therefore also examined the effect of the potent, selective 5HT2AR antagonist MDL-11939 (0, 2.5, 5, and 10 mg/kg). Figure 5c shows that 5 mg/kg of MDL-11939 produced the same locomotor inhibitory effect on WT mice as ketanserin. An additional increase in dosage (to 10 mg/kg) did not further suppress locomotor activity in either WT or Egr3−/− mice. The two-way ANOVA revealed a main effect of treatment (F(3,56)=63.8; p<0.001) and genotype (F(1,56)=4.5; p<0.01), but no treatment by genotype interaction (F(3,56)=0.7; p>0.05). While the locomotor suppression produced by MDL-11939 in WT mice was not as extreme as that of the highest doses of antipsychotics, the resistance of Egr3−/− mice to its suppressive effect was greater than to the SGAs, as MDL-11939 (up to 10 mg/kg) failed to reduce the hyperactivity of Egr3−/− mice to normal WT levels. Visual inspection of animals revealed a marked difference in activity between treated WT and Egr3−/− mice (Supplementary Video S4).

Table 2 Cloned Receptor Binding Affinities (in nM) of Study Drugs
Full size table

Egr3−/− Mice have a Deficit of 5HT2ARs in the Prefrontal Cortex

To determine whether dysfunction of 5HT2ARs may be the mechanism underlying the resistance of Egr3−/− mice to the locomotor inhibitory effects of 5HT2AR-specific agents, we conducted a radioligand binding assay to determine the level of expression of 5HT2AR in Egr3−/− mice. In the murine brain, 5HT2ARs are expressed in the frontal cortex along an anterior to posterior gradient, and show very little expression in other brain regions (Lein et al, 2007; Meltzer et al, 2010). The PFC expresses high levels of 5HT2ARs, and is also a key region implicated in schizophrenia pathogenesis in humans. We therefore dissected this region to compare receptor levels in Egr3−/− and WT using radioligand binding with [3H]ketanserin, a selective 5HT2AR ligand (Figure 6a). There was no change in the receptor binding affinity. Figure 6b shows that the maximum number of 5HT2AR/[3H]ketanserin binding sites is reduced by nearly 70% in the prefrontal cortex of Egr3−/− mice compared with WT controls (F(2,170)=14.77; p<0.001, Student's t-test).

Figure 6
figure 6

Serotonin 2A receptor (5HT2AR) levels are decreased in the prefrontal cortex (PFC) of Egr3-deficient (Egr3−/−) mice. (a) [3H]Ketanserin binding saturation curves in the frontal cortex of WT (black) and Egr3−/− (white) mice (n=8 per group). There was no change in binding affinity. (b) Maximum number of binding sites (Bmax) for [3H]ketanserin obtained from individual saturation curves. (c) Egr3−/− mice display a reduced behavioral response to 5HT2AR hallucinogenic agonist 1-(2,5-dimethoxy 4-iodophenyl)-2-amino propane (DOI). Head-twitch responses were recorded for 30 min post-administration of DOI (1 mg/kg) or vehicle (*p<0.0001, Student's t-test).

PowerPoint slide

Full size image

Egr3−/− mice also displayed a decreased behavioral response to DOI (1 mg/kg), a 5HT2AR agonist that produces a distinctive head-twitch response in WT mice (Darmani et al, 1990; Gonzalez-Maeso et al, 2003) (Figure 6c). The two-way ANOVA revealed a main effect of treatment (F(1,20)=33.1; p<0.001) and a treatment by genotype interaction (F(1,20)=9.1; p<0.01). In summary, these results demonstrate a functional deficit of membrane-bound 5HT2ARs in the PFC of Egr3−/− mice.

DISCUSSION

The aim of this study was to elucidate the mechanism underlying our previously published observation that Egr3−/− mice are resistant to the locomotor suppression produced in WT mice by clozapine, a uniquely effective antipsychotic medication that remains one of the leading treatments for schizophrenia (Gallitano-Mendel et al, 2008; Kane et al, 1988). In addition, this study simultaneously addressed our larger objective, to identify a downstream effector of Egr3. Such a gene would be a potential next step in our hypothesized biological pathway influencing schizophrenia susceptibility.

In the past decade, human genetics studies have identified numerous genes that are associated with risk to develop schizophrenia, a severe mental illness that affects 1% of the world's population. However, any individual gene is only able to account for a small percentage of illness risk (Allen et al, 2008; Owen et al, 2010). One way to unite multiple candidate genes conceptually is to identify those which act in a common biological pathway. EGR3 is an immediate-early gene transcription factor that is regulated downstream of three major proteins implicated in schizophrenia susceptibility (neuregulin 1; (Hippenmeyer et al, 2002; Jacobson et al, 2004; Stefansson et al, 2002), N-methyl D aspartate receptors (NMDARs) (Olney et al, 1999; Yamagata et al, 1994), and calcineurin (Mittelstadt and Ashwell, 1998; Yamada et al, 2007)), and was recently identified as the central gene in a network of transcription factors and microRNAs implicated in schizophrenia susceptibility (Guo et al, 2010). Moreover, EGR3 is, itself, associated with schizophrenia (Kim et al, 2010; Yamada et al, 2007), and expressed at reduced levels in post-mortem brain tissue from schizophrenia patients (Mexal et al, 2005; Yamada et al, 2007). These findings suggest the need for further investigations of a potential role for EGR3, and the biological pathway of genes in which it functions, in psychotic disorders.

Our ‘pharmacological dissection’ approach proved successful in revealing that 5HT2AR-specific antagonists parallel the activity of clozapine in suppressing the locomotor activity of WT mice at dosages that fail to reduce the activity of Egr3−/− mice below normal WT activity levels. This is not due to the basal hyperactivity of Egr3−/− mice, as FGAs haloperidol (previously reported) and chlorpromazine (Figure 2) suppress the locomotor activity of Egr3−/− mice at the same dosage as WT mice.

We hypothesized that a defect in the function of 5HT2ARs in the Egr3−/− mice could explain their differential sensitivity to 5HT2AR antagonists. Indeed, receptor binding studies using the 5HT2AR-selective ligand ketanserin revealed a nearly 70% reduction in 5HT2AR activity in the PFC of Egr3−/− mice (Figure 6a and b). This reduction of receptors corresponded with the results of a functional assay, the head-twitch response to the 5HT2AR agonist DOI (figure 6c), a drug-induced behavior that is absent in 5HT2AR knockout mice. Thus, it appears that the reduced sensitivity of Egr3−/− mice to the locomotor suppressive effects of clozapine and other SGAs may be a result of decreased levels of 5HT2ARs in the brains of the mice. Further investigation of the relative affinities of the FGAs and SGAs tested (using values from the PDSP website (Roth, 2008) and Table 2) indicated that the ratio of 5HT2AR to D2R binding affinity best correlated with the locomotor inhibitory response of Egr3−/− mice.

Findings in our animal model are notable as numerous studies have identified deficits in 5HT2AR levels in the brains of schizophrenia patients (Dean and Hayes, 1996; Erritzoe et al, 2008; Garbett et al, 2008; Hurlemann et al, 2008; Lopez-Figueroa et al, 2004; Matsumoto et al, 2005; Ngan et al, 2000; Rasmussen et al, 2010; Serretti et al, 2007). Moreover, the HTR2A gene, which encodes the 5HT2AR, is a leading candidate schizophrenia gene (Allen et al, 2008). Thus, the deficit of 5HT2AR in Egr3−/− mice suggests a possible mechanism through which EGR3 (itself a candidate schizophrenia gene) may influence susceptibility to this mental illness. Furthermore, this finding suggests that the 5HT2AR may act downstream of EGR3 in what we hypothesize to be a biological pathway of genes influencing schizophrenia risk.

Insights into the Mechanisms of SGA-Induced Locomotor Suppression

To date, the precise mechanism by which clozapine exerts its antipsychotic effects remains unclear. Similarly, the etiology of side effects, such as sedation and weight gain, are also uncertain. Despite this, the sedating effect of clozapine has frequently been attributed to antagonism of H1 histamine receptors (Casey, 1997; Mengod et al, 1996; Stahl, 2008). Our results suggest that this is not the case in C57BL/6 mice, as selective H1 antagonists fail to reduce WT locomotor activity even at the highest doses reported used in mice in the literature (Figure 1 and Table 1) (Parsons and Ganellin, 2006; Shishido et al, 1991). These findings suggest the possibility that selective antagonism of H1 receptors may not be the mechanism responsible for the sedating effect of SGAs in humans either.

However, it is possible that sedation in humans may differ from the locomotor suppression we see in mice. Species differences in the molecular regulation of psychoactive medications have been reported (Gershon et al, 2011). Further investigation in primates and humans are needed to assess whether these results translate across species. Alternatively, H1 antagonism may have a different effect in combination with drug activity at other receptors than it does alone. In fact, studies investigating low-dose administration of psychiatric medications are aimed at determining the relative influence of H1 receptors vs other receptors involved in brain activation, in the sedating characteristics of these medications (Casey (1997) and references therein).

Instead, our findings suggest that the locomotor suppressive effect of SGAs in mice may result, at least in part, from the binding of these medications to 5HT2ARs. Our data demonstrate that drugs which selectively target this receptor, ketanserin and MDL-11939 (Figure 5), parallel the effect of clozapine, and other SGAs, on Egr3−/− mice. They suppress locomotor activity in WT mice at dosages that partially or completely reverse the hyperactivity of Egr3−/− mice, but do not reduce their activity below that of vehicle-treated WT mice. However, although these agents show a similar divergence in their locomotor suppressive effects on WT and Egr3−/− mice as do the SGAs, they do not suppress the locomotor activity of WT mice to the same degree as the SGAs, which block movements in a 1 h test to nearly zero. Thus, although the reduction in PFC 5HT2ARs may be sufficient to reduce sensitivity of Egr3−/− mice to the locomotor suppressive effects of SGAs, the blockade of other receptors in combination with 5HT2ARs may be contributing to the activity-suppressing effects of SGAs in WT mice. Finally, the possibility that the 5HT2AR may contribute to the sedating effects of SGAs in humans is less surprising when one considers that 5HT2AR-specific antagonists and inverse agonists are being investigated by the pharmaceutical industry as sleep aids (Teegarden et al, 2008).

Our findings are consistent with those of McOmish and colleagues (2010), who recently reported that 5HT2AR−/− mice show the same resistance to the locomotor suppression produced in WT mice by clozapine that we previously reported in Egr3−/− mice. Using a conditional regional rescue of 5HT2AR function, they demonstrated that this response results from loss of 5HT2ARs in the cortex, and is not caused by receptor loss in the striatum. This suggests that the reduction in 5HT2ARs we have identified in the cortex of Egr3−/− mice is, indeed, responsible for their differential response to clozapine and other SGAs, compared with FGAs.

Our study does not address whether the loss of 5HT2AR expression may contribute to other phenotypes that the Egr3−/− mice display, including schizophrenia-like behavioral abnormalities (Gallitano-Mendel et al, 2007; (Gallitano-Mendel et al, 2008). However, the nearly 70% reduction in cortical 5HT2ARs does not appear to disrupt the effectiveness of clozapine, as we have previously reported that chronic clozapine is able to reverse the aggressive behavior of Egr3−/− mice (Gallitano-Mendel et al, 2008). This is consistent with a recent report by Yadav and co-workers (2011), which demonstrated that post-synaptic 5HT2ARs are not essential for clozapine's ability to reverse phencyclidine-induced disruption of sensory–motor gating in mice, an NMDAR hypofunction animal model of schizophrenia.

FGA and SGA Antipsychotics

Inspection of the binding profiles of the FGAs and SGAs, as shown in Table 2, does not reveal a single receptor that can explain the differential susceptibility of Egr3−/− mice to the locomotor suppressive actions of these two classes of drugs. Like SGAs, the FGAs also bind to 5HT2ARs. In fact, the affinity of chlorpromazine for the 5HT2AR is nearly identical to that of olanzapine and ziprasidone and, according to the PDSP source (Roth, 2008), is greater than that of clozapine. As noted earlier, the ratio of 5HT2AR to dopamine D2 receptor affinities appears to most closely align with the differential susceptibility of Egr3−/− mice to locomotor suppressive effects of these drugs. Notably, Meltzer and co-workers (1989, 2003) have hypothesized that the 5HT2AR : D2R ratio is the main characteristic that distinguishes the FGAs from SGAs.

Despite the importance of SGAs, which became first-line treatments for schizophrenia from the late 1990s to early 2000s, there is no simple experimental assay for distinguishing FGAs from SGAs. Such an assay would be beneficial for screening novel candidate molecules for antipsychotic characteristics (Geyer and Ellenbroek, 2003). One screening test has been reported, but it involves extensive behavioral training of animals followed by multiple pharmacological interventions, and is thus difficult and time-intensive (Philibin et al, 2005). Our finding that the behavioral response of Egr3−/− mice appears to distinguish SGAs from FGAs suggests that these mice may provide a rapid assay for this purpose.

Further work is needed to identify the etiology of the reduced PFC 5HT2AR binding in Egr3−/− mice. In particular, studies aimed at identifying whether the dysfunction is at the level of protein localization or translation, or gene expression, must be undertaken. These studies are challenging as antibodies against the 5HT2AR have been notoriously poor for immunohistochemical and western blot methods. While there has been some recent progress in this area, they still provide poor anatomical resolution (Weber and Andrade, 2010) and are less sensitive than radiological receptor binding assays. As Egr3 is a transcription factor, it is intriguing to hypothesize that it may directly regulate expression of the Htr2a gene. However, as an immediate-early gene, Egr3 expression is stimulus-dependent and its basal expression is low. We have found that systematic induction of Egr3 expression is necessary to identify putative target genes. These studies are beyond the scope of the current report, but are important areas for future investigation to identify the mechanism by which Egr3 influences this important receptor.