Exposure to the cytokine EGF leads to abnormal hyperactivity of pallidal GABA neurons: implications for schizophrenia and its modeling


  • Hidekazu Sotoyama,

    1. Department of Molecular Neurobiology, Brain Research Institute, Niigata University, Niigata, Japan
    Search for more papers by this author
  • Hisaaki Namba,

    1. Department of Molecular Neurobiology, Brain Research Institute, Niigata University, Niigata, Japan
    Search for more papers by this author
  • Satomi Chiken,

    1. Division of System Neurobiology, National Institute for Physiological Sciences and Department of Physiological Sciences, Graduate University for Advanced Studies, Myodaiji, Okazaki, Japan
    Search for more papers by this author
  • Atsushi Nambu,

    1. Division of System Neurobiology, National Institute for Physiological Sciences and Department of Physiological Sciences, Graduate University for Advanced Studies, Myodaiji, Okazaki, Japan
    Search for more papers by this author
  • Hiroyuki Nawa

    Corresponding author
    • Department of Molecular Neurobiology, Brain Research Institute, Niigata University, Niigata, Japan
    Search for more papers by this author

Address correspondence and reprint requests to Hiroyuki Nawa, Department of Molecular Neurobiology, Brain Research Institute, Niigata University, Asahimachi-dori 1-757, Niigata 951-8585, Japan. E-mail: hnawa@bri.niigata-u.ac.jp


Previous studies on a cytokine model for schizophrenia reveal that the hyperdopaminergic innervation and neurotransmission in the globus pallidus (GP) is involved in its behavioral impairments. Here, we further explored the physiological consequences of the GP abnormality in the indirect pathway, using the same schizophrenia model established by perinatal exposure to epidermal growth factor (EGF). Single-unit recordings revealed that the neural activity from the lateral GP was elevated in EGF-treated rats in vivo and in vitro (i.e., slice preparations), whereas the central area of the GP exhibited no significant differences. The increase in the pallidal activity was normalized by subchronic treatment with risperidone, which is known to ameliorate their behavioral deficits. We also monitored extracellular GABA concentrations in the substantia nigra, one of the targets of pallidal efferents. There was a significant increase in basal GABA levels in EGF-treated rats, whereas high potassium-evoked GABA effluxes and glutamate levels were not affected. A neurotoxic lesion in the GP of EGF-treated rats normalized GABA concentrations to control levels. Corroborating our in vivo results, GABA release from GP slices was elevated in EGF-treated animals. These findings suggest that the hyperactivity and enhanced GABA release of GP neurons represent the key pathophysiological features of this cytokine-exposure model for schizophrenia.

Abbreviations used

analysis of variance


coefficient of variation




epidermal growth factor


globus pallidus


high performance liquid chromatography


interspike interval


substantia nigra pars compacta


substantia nigra pars reticulata


subthalamic nucleus

Neurons in the globus pallidus (GP, homologous structure to the external segment of the globus pallidus) receive inhibitory afferents from the striatum, excitatory inputs from the subthalamic nucleus, and dopaminergic innervation from the substantia nigra pars compacta (Graybiel 2000; Kita 2007 for review). In turn, pallidal neurons send inhibitory outputs to the substantia nigra pars reticulata (SNr) as well as to the subthalamic nucleus (Graybiel 2000; Jaeger and Kita 2011 for review). This neural circuit is known as the indirect pathway of the basal ganglia circuit and is implicated in sensorimotor gating, motor coordination, attention, learning, and anti-psychotic pharmacology (Swerdlow et al. 2001; Takahashi et al. 2007; Nambu 2008; Qu et al. 2008, 2009). Usually, GP neurons are firing at 50–70 Hz in awake animals and the firing frequency of these neurons is modulated by glutamatergic and dopaminergic afferents (Hamada and Delong 1992; Kita et al. 2004; Chiken et al. 2008; Hadipour-Niktarash et al. 2012). A loss of dopaminergic inputs to the GP decreases the firing frequency of GP neurons, leading to the motor deficits found in Parkinson's disease (Querejeta et al. 2001; Bouali-Benazzouz et al. 2009; Chan et al. 2011; Hadipour-Niktarash et al. 2012). Conversely, dopaminergic stimulation elevates the firing frequency of GP neurons (Querejeta et al. 2001; Yasoshima et al. 2005; Hadipour-Niktarash et al. 2012), but the pathophysiological consequences of pallidal hyperactivity are poorly understood.

Hyperactivity of GP neurons is evoked by deep brain stimulation in the subthalamic nucleus (Hahn et al. 2008; Reese et al. 2011) and by striatal neurodegeneration of Huntington's disease (Temel et al. 2006). In addition to their primary effects on motor planning and execution, the deep brain stimulation and Huntington's disease sometimes elicit schizophrenia-like psycho-pathological symptoms such as delusion and hallucination (Diederich et al. 2000; Anderson and Marder 2001; Anderson and Mullins 2003; Burn and Tröster 2004; Chen et al. 2004; Smeding et al. 2006). These observations raise the hypothesis that the abnormal hyperactivity of GP neurons might be associated with delusion, hallucination, or other psychological symptoms (Graybiel 2008).

We have been exploring the immunoinflammatory hypothesis of schizophrenia through animal modeling. We find that perinatal and postnatal perturbation of epidermal growth factor (EGF)/neuregulin1/ErbB signaling evokes several behavioral endophenotypes of schizophrenia, such as deficits in sensorimotor gating, latent inhibition, social interaction, working memory, and behavioral sensitization to dopamine and methamphetamine, in adulthood (Futamura et al. 2003; Mizuno et al. 2004, 2007; Tohmi et al. 2005; Sotoyama et al. 2007; Kato et al. 2011). Our elaborated neuropathologic study reveals that, among many brain regions, GP receives dopaminergic hyperinnervation in this EGF model and plays a crucial role in the behavioral deficits (Sotoyama et al. 2011). In agreement, GP contains the highest level of EGF immunoreactivity among many brain regions of rats (Fallon et al. 1984). However, the physiological and functional changes of the pallidal neurons receiving hyperdopaminergic innervation remain to be characterized.

To address this issue, we investigated the firing activity of GP neurons in rats treated with EGF as neonates using extracellular single-unit recording. GABA release from GP neurons was also monitored using in vivo microdialysis. In addition, we prepared GP slice preparations and measured single-unit activity and GABA release to assess intrinsic and extrinsic influences on the GP activity.

Materials and methods


Male newborn Sprague–Dawley rats (Japan SLC, Hamamatsu, Japan) were housed in a plastic cage (276 × 445 × 205 mm) with a maternal rat under a 12-h light/dark cycle (lights on 8 : 00 a.m.), and allowed free access to food and water. After weaning (post-natal days 20–30), the rats were separated from littermates and grouped 2–3 rats per cage. Animal care and experiments were performed in accordance with protocols authorized by the Animal Care and Use Committee of Niigata University. Each animal (post-natal days 56–94) was only used in a single experiment. All efforts were made to minimize both the suffering and number of animals used in this study.

EGF treatment

Recombinant human EGF (Higeta-Shoyu Co, Chiba Japan) was dissolved in saline and administered daily by subcutaneous injection (0.875 μg/g) on post-natal days 2–10 (hereafter, referred to as EGF rats). Littermates received cytochrome c injections (0.875 μg/g; Sigma-Aldrich, St. Louis, MO, USA dissolved in saline) on the same schedule as above; they served as control animals (Futamura et al. 2003).

In vivo extracellular recording

Extracellular single-unit recording was performed at the age of 8–10 week under chloral hydrate anesthesia (400 mg/kg i.p.) (Pan and Walters 1988; Ruskin et al. 1999). Anesthetized rats were mounted on a stereotaxic apparatus, and their body temperature was continuously controlled to 37.0 ± 0.5°C. The skull overlying the GP was removed. A glass microelectrode filled with 0.5 M NaCl containing 2% Pontamine Sky Blue (Resistence; 15–25 MΩ) was inserted into the GP. The stereotaxic coordinates were anterior–posterior (AP) 0.8–1.2 mm posterior to bregma, lateral (L) 3.2–4.0 mm, and dorsal–ventral (DV) 5.0–7.0 mm ventral from the dural surface according to Paxinos and Watson (1998). Neuronal signals were recorded for 2–3 min and amplified using an amplifier (Axoclamp 2B; Molecular Devices, Sunnyvale, CA, USA) connected to a high gain amplifier (AVH-11; Nihon Kohden, Tokyo, Japan). Single units were constantly monitored by the window discriminator (121 Window Discriminator, World Precision Instruments, Sarasatota, FL, USA). The signals were displayed on a digital storage oscilloscope (VC-6523; Hitachi Kokusai Electric Inc., Tokyo, Japan), and transferred via a digitizer (Digidata 1200; Molecular Devices) to a computer equipped with recording software (Axoscope 1.1; Molecular Devices). During single-unit recording, electroencephalogram was also recorded from a screw anchored on the skull and amplified using an amplifier (Model DAM80; World Precision Instruments) to monitor anesthetic levels. At the end of the single-unit recording, the final recording site was marked by electrophoretic injection of Pontamine Sky Blue with negative current (20 μA) for 15 min. The mean firing frequency and several measures of firing patterns were analyzed from the initial 500 spike events of digitized recordings. For firing pattern analyses, the following parameters of interspike intervals (ISIs) were calculated in each cell; coefficient of variation (CV), kurtosis (3.0 in normal distribution), and skewness of ISIs; burst index (a ratio of the mean ISI to the mode ISI) (Hutchison et al. 1998).

A atypical anti-psychotic, risperidone (1 mg/kg; Janssen Pharmaceutical K.K., Tokyo, Japan), or saline was administered (i.p.) daily to some EGF and control rats for 14 days before recording. On day 15, in vivo extracellular single-unit recording was performed in the lateral area of the GP.

In vitro extracellular single-unit recording

The rats were anesthetized with halothane, and the brains were quickly removed and cooled for 5 min in an ice-cold slicing solution containing 120 mM choline chloride, 1.25 mM NaH2PO4, 2.5 mM KCl, 7 mM MgCl2, 0.5 mM CaCl2, 26 mM NaHCO3, 15 mM glucose, 1.3 mM ascorbic acid, 1 mM sodium pyruvate, and bubbled with 95% O2 and 5% CO2. Each brain was mounted on a vibratome (Pro7; Dosaka EM Ltd., Kyoto, Japan) and cut to 400 μm thick horizontal slices in the same slicing solution. Typically, four to five hemispheral slices were obtained from one brain. The globus pallidus was removed from the slices and incubated at 34°C for 30 min in a Krebs solution containing 124 mM NaCl, 1.0 mM NaH2PO4, 3.0 mM KCl, 1.2 mM MgCl2, 2.4 mM CaCl2, 26 mM NaHCO3, 10 mM glucose, 1.0 mM ascorbic acid, 1 mM sodium pyruvate. Following incubation at 25°C for more than 30 min, a pallidal slice was transferred to a recording chamber where Krebs solution was perfused. In some of the experiments, the perfusion solution was switched to that supplemented with 10 μM raclopride, 200 μM dopamine, and/or with 10 μM biccuculin plus (2R)-amino-5-phosphonopentanoate (APV; 50 μM) plus 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 10 μM) (all from Sigma-Aldrich). A glass microelectrode filled with 2.0 M NaCl was inserted into a lateral area of the GP at the same coordinates of the in vivo extracellular recording. Neuronal signals were amplified using an amplifier (MEZ-8301; Nihon Kohden) connected to a high gain amplifier (AVH-11; Nihon Kohden). The signals were displayed on an oscilloscope (VC-11; Nihon Kohden), and transferred via a digitizer (Digidata 1200; Molecular Devices) to a computer equipped with recording software (Clampex 7, Molecular Devices).


Rats were anesthetized with sodium pentobarbital (65 mg/kg i.p.) and mounted in a stereotaxic apparatus. The skull was exposed and a hole was drilled for unilateral implantation of a guide cannula (21 G stainless-steel pipe; length: 20 mm) into the substantia nigra pars reticulata (SNr) (stereotaxic coordinates: AP 5.4 mm posterior to bregma, L 2.8 mm, DV 6.4 mm from the dural surface, according to Paxinos and Watson 1998). After a recovery period of more than 10 days (24 days for rats that underwent the GP lesion operation), microdialysis experiments were performed under awake conditions.

A microdialysis probe was prepared in accordance with Ichikawa et al. (2001). Briefly, a hollow dialysis membrane (0.31 mm o.d., 0.22 mm i.d., AN69HF, Hospal, Meyzieu, France) was inserted through a 25-G stainless tube (0.51 mm o.d., 0.35 mm i.d, HTX-25X-12, Small parts Inc., Miami Lakes, FL, USA.), and the tip was sealed with epoxy. The length of the exposed surface for dialyzing was 2 mm. The perfusate of artificial cerebrospinal fluid containing 147 mM NaCl, 2.7 mM KCl, 1.2 mM CaCl2, 0.5 mM MgCl2, at pH 7 was delivered at a flow rate of 1 μL/min. Five hours after implantation of the probe, dialysate was collected every 20 min at a flow rate of 1 μL/min. The first five fractions were collected to determine the basal levels of glutamate and GABA. The next two fractions were collected during the perfusion of a medium containing high concentration of potassium (80 mM KCl, 69.7 mM NaCl, 1.2 mM CaCl2, 0.5 mM MgCl2, pH 7.0). After potassium perfusion for 40 min, the perfusion medium was switched back to the original medium and four fractions were collected.

Pallidal lesion

Control and EGF rats (post-natal days 56–70) were anesthetized with sodium pentobarbital (65 mg/kg i.p.). After confirming deep anesthesia, the rat was mounted on a stereotaxic apparatus with an incisor bar set at 3.3 mm below the interaural line. The skull was exposed and a hole was drilled for unilateral injection into the GP. A 30 G stainless-steel needle was placed into the left GP (stereotaxic coordinates: AP 1.0 mm posterior to bregma, L 3.2 mm, DV 6.8 mm from the dura surface, according to Paxinos and Watson 1998). Five microgram of ibotenic acid (Sigma-Aldrich; dissolved in 0.5 μL sterilized phosphate buffered saline, pH 7.5) or saline was injected at a flow rate of 0.1 μL/min, and the needle was kept in place for an additional 5 min to attenuate the leakage of ibotenic acid into the needle track. After the injection, a guide cannula for microdialysis was implanted into the SNr immediately. Twenty-four days after surgery, the microdialysis experiment was performed.

Monitoring GABA release in vitro

Brains were dissected as described above. Coronal slices (450 μm thick) were cut using a vibratome (Pro7; Dosaka EM Ltd). The region of the GP was punched out and incubated at 34°C for 60 min in Krebs solution (described above). GP slices were transferred to a 20 mm diameter mini-chamber containing 0.5 mL of Krebs solution and incubated at 34°C for 7 min. Glutamate and GABA concentrations in the supernatant were determined as described below.

High performance liquid chromatography

Glutamate and GABA concentrations were determined via the HPLC system combined with a fluorometric detector (excitation: 350 nm; emission: 450 nm; RF-10AXL, Shimazu, Kyoto, Japan). Samples were derivatized with o-phtalaldehyde and injected to a reverse phase column (CLC-ODS, 4.6 × 150 mm, Shimazu). The mobile phase (0.1 M sodium acetate, 50 mg/L EDTA-3Na, acetonitrile 15–50%, pH 6.0) was delivered at a flow rate of 0.5 mL/min.

In the assessment of GABA and glutamate releases from GP slices, the amounts of GABA and glutamate were normalized with the total amino acid content of the sample. The amount of total amino acids was determined by Flow-Injection Analysis (Hung et al. 2010). Data analysis was performed by an analysis software, Epsilon LC ver.2.34 (Bioanalytical Systems Inc., West Lafayette, IN, USA).

Histological examination

After in vivo electrophysiology or microdialysis, rats were deeply anesthetized with halothane and decapitated. Brains were quickly removed and fixed in 4% paraformaldehyde for 3 days. Fixed brains were cut into 50-μm sections using a vibratome (DTK-2000; Dosaka EM Ltd.). Each section was stained with Cresyl Violet. The location of recorded neurons and the microdialysis probe was determined under a microscope according to a stereotaxic atlas (Paxinos and Watson 1998). Animals that exhibited incorrect cannula placement were removed from the data analysis; three rats in the microdialysis experiment and zero in the in vivo recording.

Statistical analysis

When data did not fit a Gaussian distribution, data were analyzed by a non-parametric comparison of Mann–Whitney's U-test or Kruskal–Wallis test. Otherwise, a two-way analysis of variance (anova) was applied using the between subject factor of EGF treatment (two levels) and lesion (two levels) or within subject factor of time (multiple levels). Post hoc comparisons were performed by Scheffe's comparison or Fisher's Least Significant Difference (LSD). Alternatively, we used an unpaired two-tailed t-test for a simple comparison of transmitter concentrations between two experimental groups. A p-value of less than 0.05 was considered to indicate statistical significance. Statistical analyses were performed using StatView software (SAS Institute Inc., Cary, NC, USA). Values in parentheses (n) represent the number of animals used in each group.


Spontaneous neuronal activity is increased in the lateral GP in vivo

We monitored GP neuronal activity of EGF and control rats using in vivo single-unit recording in an anesthetic condition (Fig. 1). We divided the GP into three areas; the medial, central, and lateral areas (Fig. 1a). We measured neural activity in the central and lateral areas but did not in the medial region because of poor recording efficacy in this region. In the lateral area (Fig. 1b), a mean firing rate of EGF rats was 22.6 ± 1.2 Hz, which was higher than that of control rats (16.7 ± 1.2 Hz, 88 cells) (p < 0.001). In the central region (Fig. 1c); however, there was no significant difference in firing rates: 20.2 ± 1.2 Hz for controls and 23.3 ± 1.5 Hz for EGF rats (= 0.14).

Figure 1.

Elevated spontaneous neural activities of pallidal neurons in epidermal growth factor (EGF) rats. In vivo extracellular single-unit recording was performed in the globus pallidus (GP) of anesthetized EGF and control rats (five animals each). (a) The GP was divided into three regions: the lateral (lateral 3.8–4.0 mm from the midline; dark gray), central (lateral 3.2–3.8 mm; light gray), and medial (white) areas. Mean firing rates of GP neurons in the lateral (b) and central (c) areas of control and EGF rats were plotted in the probability histogram at 5-Hz intervals. Typical single-unit activity of GP neurons is displayed. The distribution of mean firing rates were compared in the lateral GP between treatment groups by Mann–Whitney U-test (= 0.002; = 121 cells for EGF, n = 88 cells for control). In the central region, however, there was no significant difference in firing rates (= 0.14; n = 96 cells for EGF, n = 119 cells for control).

Chronic risperidone treatment normalizes mean spike frequency in the lateral GP

Our previous study showed that chronic treatment of the atypical anti-psychotic risperidone ameliorates abnormal pallidal dopamine release as well as prepulse inhibition deficits of EGF-treated rats (Sotoyama et al. 2011). These results raised the question of whether chronic treatment of the anti-psychotic also normalizes the GP hyperactivity. To address this question, we performed in vivo single-unit recording in the lateral area of the GP after chronic risperidone treatment (Fig. 2a). The mean firing rate was 16.3 ± 1.0 Hz in saline-injected control rats, 24.6 ± 1.7 Hz in saline-injected EGF rats, 14.5 ± 1.3 Hz in risperidone-injected control rats and 15.7 ± 1.3 Hz in risperidone-injected EGF rats (Fig. 2b). A Kruskal–Wallis test revealed significant differences between the four groups (< 0.001). Scheffe's post hoc comparisons revealed that the enhanced neural activity of saline-injected EGF rats was normalized by the risperidone treatment (p < 0.001). There was no significant effect of risperidone on mean firing rates of control rats, however.

Figure 2.

Effects of chronic risperidone administration on pallidal neural activities. (a) Risperidone (1 mg/kg) or vehicle (saline) was administered (i.p.) to rats daily for 14 days. On day 15, in vivo extracellular single-unit recording was performed in the lateral GP of vehicle-injected and risperidone-injected epidermal growth factor (EGF) rats and vehicle-injected and risperidone-injected control rats (n = 5 rats each). (b) Mean firing frequency of individual GP neurons in control rats receiving saline (Control+Vehicle), control rats receiving risperidone (Control+Risperidone), EGF rats receiving saline (EGF+Vehicle), and EGF rats receiving risperidone (EGF+Risperidone) were plotted. For firing pattern analyses, coefficient of variation of interspike intervals (ISIs) (c), skewness of ISIs (d), and kurtosis of ISIs (e) of individual GP neurons were plotted and compared among groups by the Kruskal–Wallis test, followed by Scheffe's post hoc comparisons; *p < 0.05, **p < 0.01, ***p < 0.001. Horizontal lines represent mean values of the distributions.

Using these recording data of the four groups, we also analyzed and compared the spike patterns of individual cells among groups (Fig. 2c–e). CV, kurtosis, and skewness of ISIs in each cell were calculated and plotted. Kruskal–Wallis test revealed significant group differences in all these indices (p = 0.01~0.0001). Post hoc tests detected the common effects of risperidone on CV in both control and EGF rats (p = 0.001–0.028). Post hoc tests also detected significant increases in kurtosis of ISIs (p = 0.007–0.027), and skewness (p = 0.002–0.023) in risperidone-injected EGF rats, compared with vehicle-injected control rats as well as with risperidone-injected control rats. These results suggest that the risperidone treatment decreased the relative number of high frequency events (i.e., longs ISI), suggesting that risperidone enhanced a bursting trend of GP firing. In agreement, burst index (Hutchison et al. 1998) differed significantly among four groups (p < 0.001, a Kruskal–Wallis test); there was a marginal trend toward increasing bursts in the EGF+risperidone group (18.6 ± 9.0), compared with the EGF+Vehicle group (2.8 ± 0.5) (p = 0.10, a Scheffe's test). However, we failed to detect EGF effects on the other GP firing patterns.

Elevation of spontaneous firing in GP slices of EGF rats

Dopamine acts on GABAergic striatopallidal terminals and glutamatergic subthalamopallidal terminals and regulates GP activity (Nambu and Llinaś 1994; Kita et al. 2004; Jaeger and Kita 2011). To limit the influences of the afferent nerves, we made slice preparations from the GP tissue of EGF and control rats and monitored single-unit activity in the lateral area of the GP (Fig. 3). We obtained a similar difference to the results in vivo; the mean firing rates were significantly higher in EGF rats (10.9 ± 1.0 Hz) than in controls (8.3 ± 0.8 Hz) (p = 0.041).

Figure 3.

An increase in spontaneous neural activity in globus pallidus (GP) slice preparations. Extracellular single-unit recordings were performed in GP slice preparations (12 hemispheral slices from three rats for each group). (a) Typical single-units activity of control and epidermal growth factor (EGF) rats is displayed. (b) Distribution of the mean firing rates was displayed in the probability histogram at 2-Hz intervals. Probability distributions were compared by the Mann–Whitney U-test (n = 29 cells for EGF, n = 33 cells for control).

We initially assumed that slice preparation should eliminate the influences of all the afferent activities to limit the dopaminergic effect. With this respect, the above increase in GP slices of EGF rats was beyond our expectation. Accordingly, we further dissected the dopaminergic influences, challenging GP slices with an antagonist and an agonist for dopamine receptors. GP slices were prepared from EGF rats and treated with an antagonist for dopamine D2 receptors, raclopride. We found significant effects of raclopride on unit activity even in slices (F(1,9) = 11.2, p = 0.009), suggesting that dopaminergic terminals were still alive and active in slice preparations even they lost the cell bodies (Fig. 4a).

Figure 4.

Effects of dopamine on pallidal neural activity in slice preparations. Extracellular single-unit recordings were performed in globus pallidus (GP) slice preparations from epidermal growth factor (EGF) or normal rats. Mean unit activity was calculated at one min intervals. Percent changes of the activity were shown. (a) After a 5-min recording of pallidal basal activity, an antagonist of dopamine receptor D2 (10 μM raclopride) was perfused to pallidal slices of EGF rats for 10 min. (b) To distinguish pre-synaptic and post-synaptic effects of dopamine on GP neurons, an agonist, dopamine (DA; 200 μM) was perfused to pallidal slices of normal rats in the presence or absence of a GABA receptor blocker (10 μM bicuculline) and glutamate receptor blockers (50 μM APV and 10 μM CNQX). Relative unit activity was adjusted just before dopamine application between groups. Dopamine-triggered changes were compared between the presence and absence of the blockers (bicuculline, APV and CNQX); *p < 0.05, **p < 0.01. The effect of these blockers on the basal unit activity was also determined; ###p < 0.001, all compared by Fisher's LSD (n = 5 cells each).

The above result suggested that there were also substantial influences of GABAergic terminals and glutamatergic terminals remaining in GP slices. Thus, we tested the effects of dopamine challenge on GP activity in the presence or absence of the GABA blocker, bicuculline, and the glutamatergic inhibitors, APV and CNQX, which presumably block the residual activities of GABAergic and glutamatergic afferents in slices. The application of dopamine alone to pallidal slices from normal rats significantly elevated unit activity of GP neurons (Fig. 4b, blue). This dopamine-triggered activity increase was partially attenuated by the co-application of the GABA blocker and the glutamatergic inhibitors (significant interaction of dopamine x time; F(1,15) = 1.961, p = 0.034, post hoc comparison by Fisher's LSD; p = 0.008–0.047). However, the GABAergic and glutamatergic blockers themselves had significantly elevated the basal unit activity (Fig. 4b, purple). These results suggest that dopamine terminals survived even in slice conditions and directly or indirectly influenced GP activity.

Elevation of basal GABA release in the SNr of EGF rats

A majority of the GABAergic neurons projecting to the SNr contain parvalbumin and are distributed in the lateral area of the GP (Smith and Bolam 1989). We anticipated that the pallidal hyperactivity should result in the facilitation of GABA release in the target regions. To examine this possibility, we performed in vivo microdialysis to assess extracellular GABA and glutamate concentrations in the SNr of EGF and control rats (Fig. 5a). We found that basal GABA levels (154 ± 12.0 nM) in the dialysate of EGF rats were significantly higher than in controls (87.9 ± 12.9 nM) (p = 0.003; Fig. 5b). However, there was no significant difference in basal glutamate levels between the two groups (717 ± 135 nM for control rats, 940 ± 256 nM for EGF rats) (p = 0.458; Fig. 5c). Following high potassium perfusion, local GABA levels were elevated but peak levels were indistinguishable between EGF and control rats (F(1,10) = 1.17, p = 0.304; Fig. 5b). There was no significant difference in potassium-evoked efflux of glutamate (F(1,10) = 0.74, p = 0.41; Fig. 5c).

Figure 5.

Basal GABA levels in the substantia nigra pars reticulata (SNr) of epidermal growth factor (EGF) rats. In vivo microdialysis was performed in the substantia nigra pars reticulata (SNr) of awake EGF and control rats (n = 6 each). (a) The locations of dialysis probes were shown in coronal brain sections. The digit represents the distance from the bregma. (b) Concentrations of GABA in dialysates were determined by HPLC. A horizontal solid bar represents a high potassium-stimulation (80 mM KCl). Plots in the window show basal levels of GABA with a magnified scale. (c) Glutamate concentrations in the same dialysates were determined and plotted. **p < 0.01, compared by Fisher's LSD.

GP lesion diminishes basal GABA release in the SNr

To confirm that the GABA increase in SNr reflects the increase in GABA release from pallidonigral afferents, we made GP lesions by ibotenic acid injection and assessed their effects on local GABA concentrations in SNr of EGF and control rats (Fig. 6). Histological examination showed a massive loss of GP neurons in lesioned group compared to the sham operation group, confirming the authenticity of GP lesioning (Fig. 6b). The majority of GP neurons were eliminated (see Figure S1a, b). A two-way anova with treatment (cytochrome c vs. EGF) and lesion (sham vs. lesion) as subjects factors revealed significant main effects of treatment (F(1,28) = 6.55; p = 0.016) and lesion (F(1,28) = 16.6; = 0.003), and a significant interaction of EGF × lesion (F(1,28) = 8.91; p = 0.006) (Fig. 6c). Post hoc comparisons revealed the GP lesion markedly reduced basal GABA levels in EGF rats (p < 0.001) (229 ± 41 nM for EGF+sham operation; 48.8 ± 13.7 nM for EGF+lesion) but not in control rats (87.3 ± 14.8 nM for control+sham operation; 59.6 ± 6.9 nM control+lesion). However, the GP lesion elevated local glutamate concentrations (F(1,28) = 0.52, p = 0.47 for EGF; F(1,28) = 10.8, p = 0.003 for lesion; F(1,28) = 0.006, p = 0.94 for EGF × lesion) (Fig. 6d). Thus, the nigral increase in GABA levels in EGF rats was ascribed to the enhanced activity of pallidonigal afferents.

Figure 6.

Effects of globus pallidus (GP) lesions on GABA and glutamate levels in the substantia nigra pars reticulata (a) The locations of dialysis probes were shown in coronal sections. (b) Neuropathology of the GP after saline (sham) or ibotenic acid (lesion) injection was examined with the Nissle staining (see also Figure S1 for the entire GP region). Scale bar = 100 μm. (c) The effects of the GP lesion on basal GABA levels in the SNr of control (CONT) and EGF rats. (d) The effects of the GP lesion on basal glutamate levels in the SNr. Box and whisker indicate mean ± SEM (n = 8 for each sham operation group, n = 5–7 each lesioned group. *< 0.05, compared by Fisher's LSD.

Enhanced GABA release from GP slices

A majority of GP neurons contain GABA as a neurotransmitter (Smith and Bolam 1989; Parent and Hazrati 1995). Therefore, we assumed that the increase in GP activity results in local GABA release in EGF rats. We made slice preparations from the GP of EGF and control rats to measure GABA release in vitro. GABA release from GP slices of EGF rats was 4.1 ± 0.4 pmol/nmol total amino acid, which was significantly higher than that of control rats (2.8 ± 0.4 pmol/nmol total amino acid, p = 0.033; Fig 7a). The amount of glutamate release was also measured in the same medium but showed no significant difference between control and EGF rats (p = 0.53; Fig. 7b).

Figure 7.

Elevated GABA release from globus pallidus (GP) slices of epidermal growth factor (EGF) rats. Four hemispheral slices were made from the GP of each rats (six rats for each group). Two slices were placed in a chamber at 34°C for 7 min and allowed to release transmitters to the medium. (a) GABA concentrations in the medium were measured by HPLC (n = 12 for each group). (b) Glutamate concentrations were determined in the same medium (n = 12 for each group). *p < 0.05 by unpaired two-tailed t-test.


Based on the immunoinflammatory hypothesis for schizophrenia, the present animal model was established by perinatal exposure of rat pups to the cytokine EGF (Nawa et al. 2000; Futamura et al. 2003; Meyer et al. 2009). Rats and mice challenged with EGF as neonates display various behavioral endophenotypes relevant to schizophrenia at the post-pubertal stage (Futamura et al. 2003; Mizuno et al. 2004). Our previous study has supplemented the behavioral data with neuropathological information; for instance, some of the behavioral impairments can be attributed to the hyperdopaminergic pathology of the GP in this animal model (Sotoyama et al. 2011). In this study, we further explored the pathophysiological influence of the hyperdopaminergic state on the spontaneous neural activity of GP neurons. In particular, we distinguished the lateral and central areas of GP, which provide distinct GABAergic projections. The lateral region of the GP but not its medial/central regions preferentially send efferents to the SNr (Smith and Bolam 1989; Groenewegen and Berendse 1990; Shammah-Lagnado et al. 1996; Connelly et al. 2010; Cooper and Stanford 2002).

The present analysis of this animal model revealed the following physiological characteristics of the GP neurons in EGF rats. (i) Neurons located in the lateral area of the GP exhibited higher firing rates in both in vivo and slice preparations compared with those of control rats, (ii) The elevation of firing rates was ameliorated by chronic treatment with risperidone, (iii) Nigral GABA concentrations in EGF rats were higher than those in controls but were normalized by GP lesion, (iv) GABA release from GP slice preparations was also enhanced in the EGF model. These results indicate that perinatal exposure to the cytokine EGF results in persistent enhancement of the activity of the pallidonigral GABA neurons, which is normalized by the anti-psychotic risperidone. As local GABA concentrations in the SNr should represent not only GABA release from GP afferents but also the local release from the nigral GABA neurons and from striatonigral terminals of the direct pathway, we cannot rule out that the observed increase in nigral GABA concentrations, at least in part, included GABA releases from nerve terminals other than GP efferents and that the GP lesion indirectly altered nigral GABA release.

Local infusion of the D2 receptor agonist quinpirole into the GP increases the firing rates of GP neurons (Querejeta et al. 2001; Hadipour-Niktarash et al. 2012). The present results confirm this report. Given that the hyperdopaminergic innervation of the GP represents one of the pathological features of EGF rats, these observations raise the hypothesis that the hyperdopaminergic innervation should result in increased firing and function of GP neurons. Indeed, in this study, we verified the enhancement of GP activity and GABA release both in vivo and in vitro. However, the in vitro results from slice preparations appear to be controversial against the above explanation of the in vivo phenomena because all afferent nerves were cut in the slice condition. In contrast to our assumption, however, our pharmacological experiments suggest that dopamine terminals were alive in slice preparations and also acted on GP neurons to elevate their firing frequency (Nambu and Llinaś 1994).

Our previous reports have revealed that the sensorimotor gating deficit and anti-cataleptic feature of the EGF model are attributed to hyperdopaminergic neural transmission in the GP (Sotoyama et al. 2011). Although dysfunction of the GP is often implicated in motor deficits, we failed to detect any apparent impairment in locomotor functions in this model (Futamura et al. 2003; Mizuno et al. 2004; Tohmi et al. 2005). The hyperdopaminergic innervation is limited to the lateral area of the GP in this model (Sotoyama et al. 2011). Kodsi and Swerdlow (1995) also reported that this subregion of the GP was responsible for the impairment of sensorimotor gating. In agreement with these findings, this study revealed that the elevation of spontaneous firing was also limited to the lateral area of the GP. Therefore, this coincidence indicates the possibility that neural hyperactivity in the specific GP subregion might alter the neurocognitive traits of EGF rats without influencing their motor functions. However, this circumstantial evidence should be ascertained in the future with the elaborated activity manipulations of individual GP subregions.

Morphological and functional alterations of the GP are observed in patients with schizophrenia and implicated in the development of this illness. Magnetic resonance imaging (MRI) studies suggest neuropathological abnormalities in pallidal size as well as in myelination in this brain region (Hokama et al. 1995; Mamah et al. 2007; Goldman et al. 2008; Hashimoto et al. 2009). In addition, PET studies reveal that blood flow is elevated in the GP of schizophrenia patients (Early et al. 1987; Friston et al. 1992; Galeno et al. 2004). A post-mortem study indicates that schizophrenia involves a pallidal increase in D1 : D2 heterodimer-dependent affinity to dopamine (Perreault et al. 2010). These morphological and functional abnormalities of the GP are correlated with the severity of their psychopathology as well (Galeno et al. 2004; Spinks et al. 2005). These results from schizophrenia patients appear to corroborate the present pathophysiological findings from the animal model for this illness. All these observations may indicate the pathophysiological contribution of the hyperactivity of the GP to schizophrenia. Together with the given dopamine D2 receptor sensitivity of this region (Okubo et al. 1999; Willeit et al. 2006), these findings suggest that the GP might play one of key roles in schizophrenia pathology and pharmacology (Simpson et al. 1992; Perez-Costas et al. 2010; Sotoyama et al. 2011). Thus, the pathophysiological implication of the indirect pathway should be re-evaluated in individual neurobehavioral deficits relevant to schizophrenia.


Recombinant human EGF was kindly provided from Higeta-Shoyu Co. Ltd. This study was supported by MEXT KAKENHI (No.24116010), JSPS KAKENHI (No. 22300107, No. 23500464), Core Research for Evolutional Science and Technology from the JST Corporation and NIPS Collaborative Project. The funders and the company had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors have declared that no competing interests exist.