Neuregulin-1 (NRG1) is implicated in the etiology or pathology of schizophrenia, although its biological roles in this illness are not fully understood. Human midbrain dopaminergic neurons highly express NRG1 receptors (ErbB4). To test its neuropathological role in the neurodevelopmental hypothesis of schizophrenia, we administered type-1 NRG1 protein to neonatal mice and evaluated the immediate and subsequent effects on dopaminergic neurons and their associated behaviors. Peripheral NRG1 administration activated midbrain ErbB4 and elevated the expression, phosphorylation and enzyme activity of tyrosine hydroxylase (TH), which ultimately increased dopamine levels. The hyperdopaminergic state was sustained in the medial prefrontal cortex after puberty. There were marked increases in dopaminergic terminals and TH levels. In agreement, higher amounts of dopamine were released from this brain region of NRG1-treated mice following high potassium stimulation. Furthermore, NRG1-treated mice exhibited behavioral impairments in prepulse inhibition, latent inhibition, social behaviors and hypersensitivity to methamphetamine. However, there were no gross abnormalities in brain structures or other phenotypic features of neurons and glial cells. Collectively, our findings provide novel insights into neurotrophic contribution of NRG1 to dopaminergic maldevelopment and schizophrenia pathogenesis.

Neurodevelopmental deficits are considered to be the key features of schizophrenia, which is a multifactorial disease involving environmental factors/insults and genetic predispositions, such as a genetic polymorphism of the neurotrophic factor neuregulin-1 (NRG1). NRG1 and its receptor (ErbB4) were identified as susceptibility genes for schizophrenia.^{1, 2} Changes in the expression levels of NRG1 splicing isoforms and ErbB4 protein are also found in post-mortem brains and peripheral blood cells of schizophrenia patients,^{3, 4, 5, 6} although the pathophysiological contribution of abnormal NRG1/ErbB4 signaling to schizophrenia is largely unresolved. In this neurodevelopmental hypothesis of schizophrenia, the environmental factors include maternal infection, abnormal delivery and neonatal hypoxia, which presumably interact with the risk genes of schizophrenia.^{7, 8} For example, schizophrenia-related single-nucleotide polymorphisms (SNPs) of the NRG1 gene are often located in its promoter regions and positively regulate gene transcription.^{1, 5} NRG1 expression is induced by adult ischemic and traumatic brain injury, as well as by neonatal hypoxia.^{9, 10, 11} Thus, it is possible that the more abnormal expression of NRG1 is induced in human embryos or neonates carrying these SNPs by these environmental insults, the more severely this factor might impair brain development to increase the risk of schizophrenia.

Neuregulin-1 is one of the neurodevelopmental regulators that are involved in neuronal migration, axon pathway finding, myelination and synaptogenesis.^{12, 13, 14, 15} Thus, the abnormality in its expression can be implicated in the neurodevelopmental hypothesis mentioned above. Accordingly, various exons of the NRG1 genome have been disrupted by homologous recombination in mice and their neurobehavioral traits have been investigated.^{1, 16, 17, 18, 19} In adults, the mutant mice of NRG1 variants often exhibit schizophrenia-associated behavioral abnormalities. As hypomorphic or hypermorphic NRG1 signals persists throughout life in these genetic mutants (that is, is not temporally controlled), the evaluation of these models is challenging in regard to the neurodevelopmental hypothesis.

Recently, we described the localization of ErbB4 mRNA in the midbrain dopaminergic neurons in mice and primates including humans.^{20, 21} Our in situ hybridization detected high levels of ErbB4 mRNA signals in almost all midbrain dopaminergic neurons. In particular, the expression is higher from the late embryonic stage to neonatal stage when these neurons are vigorously developing.^{20, 22} However, the nature of NRG1 activity on dopaminergic development or function is still poorly understood.²³

In this study, we have designed an experimental protocol based on the neurodevelopmental hypothesis of schizophrenia^{24, 25, 26} and assessed the pathological roles of excess NRG1 signals on dopaminergic neurons and their functions during and after development.²⁷ As upregulation of type-1 NRG1 expression is reported in schizophrenia brain pathology,^{4, 5} we selected this isoform of NRG1 protein and administered it to the periphery of mouse pups. The immediate and delayed impact of NRG1 treatment on developing dopaminergic neurons was analyzed using neurochemical and anatomical approaches. Furthermore, we discuss the use of NRG1-treated mice as a model for schizophrenia and compare it with other genetic mutant mice of this gene.

Generation of recombinant NRG1β1 protein

A cDNA for the ectodomain (46–634 nucleotides) of mouse NRG1β1²⁸ was subcloned into the pET-22b (+) vector (Novagen, Madison, WI, USA) and expressed as histidine-tagged recombinant protein in Escherichia coli (BL21 DE3, Novagen). Inclusion bodies were denatured and solubilized by 6 M guanidine–HCl, and then NRG1β1 was purified with HiTrap Chelating HP column (GE Healthcare Bio-Science AB, Uppsala, Sweden). Denatured NRG1β1 protein was refolded by gradual removal of guanidine–HCl by means of stepwise dialysis in the presence of L-arginine and oxidized glutathione.²⁹ We purified a biologically active form of NRG1β1 protein with cation exchange chromatography (HiTrap-CM-FF, GE Healthcare). The peak fractions that induced ErbB4 phosphorylation in cultured neocortical neurons were used in this study (Supplementary Figure S1). The full mature form of recombinant NRG1β1 protein carries the tag sequence of six histidine residues at its carboxyl terminal end and has calculated molecular weight of 25 400 Da. Alternatively, we obtained the core epidermal growth factor domain of human NRG1β1 (eNRG1; molecular weight 7500 Da, PeproTech EC, London, UK), which is an artificial product common to all NRG1 splice variants.

Animals and drug treatment

Pregnant C57BL/6NCrj mice were purchased from Nihon Charles River (Kanagawa, Japan), and their newborn pups were used in the following experiments. NRG1β1 or eNRG1 (both 1.0 μg g^–1 body weight) was administered subcutaneously daily to half of the littermates during postnatal days (PNDs) 2–10.³⁰ Control littermates received an injection of phosphate-buffered saline (vehicle) of the same volume. The given dose of NRG1β1 was the highest one that did not produce growth retardation in neonatal mice. A dose of NRG1β1 (3.0 μg g^–1 body weight) significantly attenuated body weight gain in the postnatal period (88.0±1.3%, compared with control mice).

Risperidon (Risperdal, Janssen Pharmaceutical KK, Tokyo, Japan) was daily administered (1.0 μg g^–1) intraperitoneally on PNDs 42–70 to induce the chronic medication state of patients.^{31, 32, 33} The same volume of physiological saline was administered as a control. Behavioral testing was conducted 24 h after the final antipsychotic administration to minimize sedative effects of risperidon. All the animal procedures were approved by the Animal Use and Care Committee of Niigata University and performed in accordance with the National Institute of Health (NIH) guideline (USA).

In situ hybridization and immunohistochemistry

Mice were anesthetized with halothane (Takeda Pharmaceutical, Osaka, Japan) and transcardially perfused with 4% paraformaldehyde. Brains were immersed in 30% sucrose and embedded in OCT compound (Sakura Finetek, Torrance, CA, USA). Coronal sections (20- to 40-μm thick) were prepared for in situ hybridization and immunohistochemistry. Alternatively, histopathological examination was performed by Klüver–Barrera stain (see Supplementary Materials and Methods).

For in situ hybridization, sections were hybridized with digoxigenin (DIG)-labeled antisense cRNA probe to ErbB4 mRNA (GenBank: NM_010154. 429–1042 nucleotides) and then with alkaline phosphatase-conjugated anti-DIG antibody as fully described previously.²⁰ For immunostaining, alternatively, sections were incubated with anti-c-fos (1:20000, Calbiochem, La Jolla, CA, USA) or anti-TH (1:1000, Millipore, Bedford, MA, USA) antibodies, followed by the biotinylated anti-rabbit immunoglobulin antibody (1:200, Vector Laboratories, Burlingame, CA, USA). The detection of primary antibodies or injected biotinylated NRG1β1³⁴ was performed with the conventional peroxidase-conjugated avidin complexes. To confirm the specificity of the avidin/biotin reaction, some of the sections were pretreated with the avidin/biotin blocking agent (Vector Laboratories). Immunoreactivity was observed using an all-in-one microscope (BZ-9000, Keyence, Osaka, Japan) and a BZ-Analyzer (Keyence).

Immunoprecipitation and immunoblotting

For immunoprecipitation, whole brain or midbrain of mice (PND 2) was homogenized in RIPA buffer (50 mM Tris–HCl buffer pH 7.4 plus 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 5 mM EDTA, 1 mM Na₃VO₄, and 1 mM NaF) containing protease inhibitor cocktail (Complete Mini, Roche, Mannheim, Germany). Protein lysate (2 mg protein) was then incubated with the anti-ErbB4 antibody (2 μg) overnight. The antigen–antibody complex was recovered with Protein G Sepharose beads (GE Healthcare) and subjected to immunoblotting as described below.

Brain tissues were homogenized in the sample lysis buffer (62.5 mM Tris–HCl pH 6.8, 2% SDS, 0.5% NP-40, 5 mM EDTA) plus the protease inhibitor cocktail. Protein samples (5–50 μg per lane) were separated by SDS–polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Membranes were probed with antibodies directed against phosphotyrosine (1:1000, Millipore), ErbB4 (1:1000, Santa Cruz Biotechnology, Santa Cruz, CA, USA), tyrosine hydroxylase (TH) (1:1000, Millipore), dopamine-β-hydroxylase (DBH) (1:500, Millipore), dopamine transporter (DAT) (1:1000, Millipore), vesicular monoamine transporter 2 (vMAT2) (1:1000, Millipore), D1 dopamine receptor (D1DR) (1:500, Santa Cruz Biotechnology), D2 dopamine receptor (D2DR) (1:250, Millipore), catechol-O-methyltransferase (COMT) (1:8000, BD Transduction Laboratories, Lexington, KY, USA), norepinephrine/noradrenaline transporter (NET) (1:500, Millipore) and β-actin (1:4000, Millipore). Alternatively, immunoblots were probed with antibodies directed against glutamatergic, GABAergic, glial markers and phospho/nonphospho-TrkB proteins (see Supplementary Materials and Methods). Immunoreactivity on membranes was detected by peroxidase-conjugated anti-immunoglobulin antibodies followed by chemiluminescence reaction combined with X-ray film exposure (ECL kit, GE Healthcare).

Behavioral testing

All behavioral tests were performed at PNDs 56–84. Spontaneous locomotor activity, acoustic startle response and prepulse inhibition (PPI) were measured as fully described previously.³⁵ The test paradigm of context- and tone-dependent fear learning was performed in a conditioning chamber and different test chamber (both; 10 L × 10 W × 10 H cm box; Obaraika, Tokyo, Japan).^{35, 36} Freezing behavior was automatically monitored by a video camera during all sessions and analyzed by imaging software (Obaraika). The latent inhibition test was performed with the same conditioning and test chambers.³⁷ For auditory brainstem-evoked response testing, social interaction and further details of individual behavioral tests, see Supplementary Materials and Methods.

Methamphetamine challenge

The effects of methamphetamine (MAP) were monitored with an automated locomotor activity monitor (Med Associates, St Albans, VT, USA). Mice were placed in an activity chamber, and their horizontal activities were recorded at 5-min intervals. Mice were first habituated to the apparatus for 60 min and then challenged by MAP (Dainippon-Sumitomo Pharmaceuticals, Osaka, Japan, 1.0 or 2.0 μg g^–1, intraperitoneally) or saline. For quantification of c-fos-positive cells, mice were fixed as described above 2 h after a single injection of MAP. We quantified the number of c-fos-positive cells in digital microscopic images using the NIH Image cell-counting system (ver. 1.61).³⁸

Surgery and microdialysis

Mice were anesthetized with pentobarbital (50 mg kg^–1, Somnopentyl, Schering Plough Animal Health, Kenilworth, NJ, USA) and mounted on a stereotaxic frame. Stainless guide cannula (AG-4, Eicom, Kyoto, Japan) with a dummy probe (AD-4, Eicom) was placed in the medial prefrontal cortex (mpFC, equivalent to prelimbic cortex) (coordinates: anterior +2.0 mm, lateral +0.5 mm, ventral −6.0 mm relative to the bregma). A probe was perfused with artificial cerebrospinal fluid (ACSF: 147 mM NaCl, 2.7 mM KCl, 1.2 mM CaCl₂, 0.5 mM MgCl₂) or ACSF containing high potassium (69.7 mM NaCl, 80 mM KCl, 1.2 mM CaCl₂, 0.5 mM MgCl₂) at the flow rate of 0.7 μl min^–1. Dialysate samples (20 μl) were collected every 30 min.

Quantification of monoamines, their metabolites and L-DOPA

The enzymatic activity of TH was assessed by monitoring the production of 3,4-dihydroxy-L-phenylalanine (L-DOPA) from tyrosine using high-performance liquid chromatography (HPLC) equipped with an electrochemical detector as fully described previously.³⁹ Contents of dopamine, dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA) and L-DOPA were determined by HPLC using a C18 column (model CA-5ODS, 4.6 × 150 mm, Eicom).³⁹ Dialysis samples were directly applied onto the HPLC system equipped with BDS Hypersil C18 column (1.0 × 100 mm, Keystone Scientific, Bellefonte, PA, USA). For details, see Supplementary Materials and Methods.

Statistical analysis

Statistical analyses were performed using SPSS 11.0 (SPSS Co., Tokyo, Japan). As the initial analyses of variance of behavioral data yielded no significant outcomes involving the gender variable, the data of the two genders were combined for final analyses. Fisher's least significant difference post hoc analysis was used to detect differences of absolute behavioral values. Alternatively, univariate data in two groups were subjected to unpaired two-tailed t-test. A P-value <0.05 was regarded as statistically significant.

Peripheral NRG1β1 crosses BBB and reaches the midbrain of mouse neonates

As the blood–brain barrier (BBB) of neonatal mice is not fully developed and may allow cytokines to penetrate into the brain,^{30, 40} we tested the BBB permeability of the full-length mature type-1 NRG1β1 (hereafter, referred to as NRG1β1) in mouse pups. We administered NRG1β1 (1.0 μg g^–1, subcutaneously) to PND 2 mice and examined the phosphorylation of NRG1 receptors (ErbB4). NRG1 administration increased the immunoreactivity for phospho-ErbB4 in the whole brain as well as in the midbrain (Figures 1a and b). The maximal activation of ErbB4 was obtained ∼3 h after injection presumably because of slow penetration rates across BBB as seen with other cytokines.³⁰ By injecting the biotinylated form of NRG1β1, we confirmed its penetration across the neonatal BBB. Significant levels of biotin signal were detected in the intracellular space as well as on cell surfaces in the midbrain (Figure 1d), potentially representing endocytosis of the ligand-bound ErbB4 receptors.⁴¹ This region contained TH-positive cells and ErbB4 mRNA (Figures 1e and f). Biotin signals were also detected in other brain regions (data not shown) but not when sections were pretreated with an avidin/biotin blocking agent (Figure 1g). The control sections prepared from vehicle-treated animals failed to exhibit a biotin signal (Figure1h). These results suggest that, during neonatal and possibly during perinatal stages, NRG1 circulating in the periphery can reach the midbrain and activate ErbB4 receptors of dopaminergic neurons.

Penetration of administered neuregulin-1 (NRG1)β1 through the blood–brain barrier. (a) The whole brain lysates were prepared at 0, 1, 3, 6 and 12 h after subcutaneous injection of NRG1β1 to neonatal mice (postnatal day (PND) 2), immunoprecipitated (IP) with the anti-ErbB4 antibody and subjected to immunoblotting (IB) with anti-phosphotyrosine (anti-PY) or anti-ErbB4 antibodies. (b) ErbB4 phosphorylation in the midbrain was similarly examined 3 h after NRG1β1 treatment (N=4 mice per group). Distributions of biotinylated NRG1β1 in the midbrain region of the enclosed area in (c) were examined with the avidin/biotinylated horseradish peroxidase complex. (d) Brain section was prepared from biotinylated-NRG1β1-injected mice. Adjoining serial sections were stained with (e) the anti-tyrosine hydroxylase (TH) antibody or (f) an in situ hybridization probe to ErbB4 mRNA. (g) Section prepared from biotinylated-NRG1β1-injected mice was pretreated with the avidin/biotin blocking reagent. (h) Brain section was prepared from vehicle-injected mice. fr, fasciculus retroglexus; a landmark for the midbrain; VTA, ventral tegmental area; SNc, substantia nigra pars compacta. Scale bar: 200 μm.

Full size image

Effects of neonatal NRG1β1 treatment on developing dopaminergic system

ErbB4, the receptor for NRG1, is expressed exclusively by the midbrain dopamine neurons.²⁰ To study the developmental effects of exogenous NRG1 on these neurons, we repeatedly administered NRG1β1 to mouse neonates (subcutaneously, PNDs 2–10). We then determined the protein expression, phosphorylation and enzyme activity of TH, a rate-limiting enzyme of dopamine and noradrenaline synthesis. We found a significant increase in TH protein levels in the whole frontal cortex (FC) (P<0.05) but not in the striatum of NRG1β1-treated mice at PND 11 (Figures 2a and b). As FC receives both dopaminergic and noradrenergic innervations, we also examined the protein levels of DBH, the enzyme that converts dopamine to noradrenaline. There was no significant change in DBH levels in FC, suggesting limited effects of NRG1β1 on noradrenergic neurons. Although NRG1β1 treatment did not affect TH levels in the striatum, there was a significant increase in ser-40 phosphorylation of this enzyme (P<0.05) (Figure 2b), a process known to elevate the enzyme activity of TH.⁴² We also tested the core epidermal growth factor domain peptide of NRG1β1 (eNRG1, common for all splice variants) as a positive control and detected similar increases in TH and its phosphorylation (Supplementary Figure 2).

Effects of neonatal neuregulin-1 (NRG1)β1 treatment on developing dopaminergic neurons. NRG1β1 (1.0 μg g^–1) or vehicle (control) was administered (subcutaneously) daily to mouse pups during postnatal days (PNDs) 2–10, and effects on dopaminergic systems were evaluated. (a) Protein levels of tyrosine hydroxylase (TH) and dopamine-β-hydroxylase (DBH) in the frontal cortex (FC) were analyzed by immunoblotting and quantified by densitometric analysis, and normalized to β-actin levels (N=6 mice per group). (b) The phosphorylation (ser-40 and ser-31) levels of TH in the striatum were analyzed by immunoblotting and are presented as the ratio of phospho-TH immunoreactivity to total TH levels (N=6–7 mice per group). (c) Enzyme activity of TH was analyzed in FC and the striatum (N=7–11 mice per group). (d) The dopamine contents were measured in FC and the striatum on PND 11. Data are expressed as mean±s.e.m (N=7–8 mice per group). ^*P<0.05, ^**P<0.01, by unpaired two-tailed t-test.

Full size image

We found that NRG1β1 treatment in neonates significantly increased the enzyme activity of TH in the striatum and FC (both P<0.05) (Figure 2c). In parallel, dopamine contents in FC and the striatum were elevated in NRG1β1-treated mice at PND 11 (FC: P<0.05, striatum: P<0.01) (Figure 2d). However, there were no differences in the dopamine metabolites, DOPAC and HVA, or in noradrenaline (DOPAC: 95.7±3.0%; HVA: 107.2±1.9%; noradrenaline: 94.1±7.1% of control). These results suggest that neonatal treatment with NRG1β1 promotes aberrant phenotypic development of midbrain dopamine neurons during neonatal and postnatal stages.

We also examined NRG1β1 effects on mouse physical development. There were no effects on body weight at PNDs 11 and 56, but there was a slight acceleration of eyelid opening and tooth eruption in NRG1β1-treated mice (Supplementary Tables S1 and S2). Thus, the transient exogenous supply of NRG1β1 seems to produce limited influences on physical indices in mouse development.

Neonatal NRG1β1 treatment induces dopaminergic hyperinnervation in adult FC

To estimate the long-term effect of neonatal NRG1β1 treatment on the dopaminergic system, we measured the levels of dopamine and its metabolites in FC, the nucleus accumbens (NAc) and striatum at the adult stage as well (PNDs 56–70). We found significant increases in dopamine metabolites, DOPAC and HVA, in FC (DOPAC: 130.0±14.0%, P<0.05; HVA: 131.0±13.5% of control, P<0.01, N=10–12 mice per group). However, there was no significant effect on dopamine in FC (105.6±6.9% of control) or on dopamine and its metabolites in the other regions (dopamine, 104.0±7.7%; DOPAC, 101.1±8.0; HVA, 100.1±8.0% of control in NAc, N=9 mice per group; and dopamine, 110.8±6.6%; DOPAC, 106.9±8.7%; HVA, 105.5±4.2% of control in the striatum, N=10−12 mice per group). To evaluate morphological influences, we marked the dopaminergic somata and fibers with TH immunostaining and examined them in serial coronal sections of FC and the midbrain at the adult stage. Denser or thicker axon terminals appeared to be distributed in the deep cortical layers of mpFC (i.e. prelimbic cortex) of NRG1β1-treated mice (Figures 3a–f). In contrast, there were no apparent differences in the frequency and arborization of these axons in other subregions of FC or in the midbrain (Supplementary Figure S3).

Neonatal exposure to neuregulin-1 (NRG1)β1 results in persistent increases in tyrosine hydroxylase (TH)-positive terminals and TH protein levels in adult medial prefrontal cortex (mpFC). Serial coronal sections of FC were prepared from vehicle-treated control and NRG1β1-treated mice and immunostained with the anti-TH antibody. TH-immunoreactive fibers in mpFC of vehicle-treated control (a, c and e) and NRG1β1-treated mice (b, d and f) are shown at the positions (a, b) +2.22 mm, (c, d) +1.98 mm and (e, f) +1.70 mm, all from the bregma. (N=3 mice per group). Scale bar: 100 μm. TH protein levels in mpFC, whole FC (FC), the nucleus accumbens (NAc) and striatum were measured by immunoblotting at the adult stage. (g) Protein levels of TH and dopamine-β-hydroxylase (DBH) in the mpFC were analyzed by immunoblotting and quantified by densitometric analysis and normalized to β-actin levels (N=6 mice per group). (h) Similarly, protein levels of TH in whole FC (FC), NAc and the striatum were analyzed (N=5–6 mice per group). ^***P<0.001, by unpaired two-tailed t-test.

Full size image

To ascertain that neonatal NRG1β1 treatment resulted in the dopaminergic hyperinnervation of mpFC at the adult stage, we examined TH protein levels with immunoblotting. In agreement, there was a significant increase in TH protein levels in the mpFC (131.0±4.3% of control, P<0.001) (Figure 3g). There was no change in DBH levels in the mpFC of NRG1β1-treated mice. This TH increase was not manifested by immunoblotting for the whole FC, NAc or striatum (Figure 3h).

Higher levels of dopamine release in the NRG1β1-treated mice verify their hyperdopaminergic states

The increases in dopamine metabolites and TH protein levels may lead to enhanced dopamine transmission at the adult stage. To test this hypothesis, we carried out in vivo microdialysis in mpFC. Although neonatal NRG1β1 treatment had no effect on the basal levels of extracellular dopamine (Figure 4a), potassium depolarization evoked higher amounts of dopamine release in mpFC of NRG1β1-treated mice (NRG1 treatment, F_(1,12)=4.00, P<0.05) (Figure 4b). The reason underlying failure in detecting the difference in basal dopamine release awaits further investigation. We also examined molecular markers related to dopamine transmission; DAT, vMAT2, D1DR, D2DR, COMT and NET. We did not detect significant changes in these proteins in mpFC, NAc or the striatum of NRG1β1-treated mice (Figures 4c and d).

Analysis of neonatal neuregulin-1 (NRG1)β1 treatment effects on adult dopamine release and on neurochemical markers of dopaminergic neurons. Neonatal mice were treated with NRG1β1 or vehicle (control) as described in Figure 2. At the adult stage (postnatal days (PNDs) 63–70), microdialysis study was carried out in the medial prefrontal cortex (mpFC). (a) Basal extracellular levels of dopamine were monitored for 150 min (N=7 mice per group). (b) Dopamine release was evoked by perfusion of 80 mM KCl over 90 min (solid bar) and monitored for 270 min. Data represent relative dopamine levels in 30-min fractions (% of baseline, mean±s.e.m, N=7 mice per group). ^*P<0.05, ^**P<0.001, by Fisher's least significant difference. Protein extracts were prepared at the adult stage (PNDs 56–84) from mpFC, the nucleus accumbens (NAc) and striatum of mice that were neonatally treated with NRG1β1 or vehicle (control) and subjected to immunoblotting with antibodies directed against the indicated dopaminergic markers. (c) Typical immunoreactive signals of two mpFC samples were displayed. (d) Immunoreactivities were measured by densitometric analysis and normalized to β-actin levels (N=5–7 mice per group). Relative levels of the protein markers in NRG1β1-treated mice are presented (% of control; mean±s.e.m). ND, not determined. Note: There were no significant differences in all pairs.

Full size image

In addition to the effects on the dopaminergic system, neonatal exposure to NRG1β1 might also influence the development of gamma-aminobutyric acid (GABA) neurons, glutamatergic synapses, glial cells or other neurotrophic signaling.^{12, 13, 43, 44, 45} Using the conventional pathological staining as well as immunoblotting, we examined brain structures and neuronal/glial markers of NRG1β1-treated mice in the adult stage (PND 56). Klüver–Barrera staining revealed that there were no apparent structural abnormalities in adult NRG1β1-treated mice (Supplementary Figure S4). In addition, there were no significant differences in the protein levels of the glutamatergic, GABAergic and glial markers (Supplementary Figure S5), and TrkB phosphorylation (Supplementary Figure S6) between NRG1β1-treated and vehicle-treated control mice.

Neonatal NRG1β1 treatment enhances behavioral and neurochemical sensitivity to MAP

We tested whether neonatal NRG1β1 treatment alters acute responsiveness to MAP in this study. A systemic challenge of MAP (1.0 μg g^–1, intraperitoneally) increased locomotor activity in both NRG1β1-treated and control groups (MAP, F_(1, 36)=16.30, P<0.001). However, the magnitude of MAP-induced locomotor activity was significantly higher in NRG1β1-treated group than the control group (NRG1 treatment, F_(1,36)=5.14, P<0.05; NRG1 treatment × MAP, F_(1, 36)=4.34, P<0.05) (Figure 5a). The acceleration of MAP sensitivity by neonatal NRG1β1 treatment seemed to depend on the MAP dose (MAP dose, F_(1, 54)=162, P<0.001). At a dose of 2.0 μg g^–1 of MAP, the effect of NRG1β1 treatment became less apparent (P=0.052) (Figure 5b). Thus, these results indicate that mice neonatally exposed to NRG1β1 exhibited higher sensitivity to the lower dose of MAP.

Neonatal exposure to neuregulin-1 (NRG1)β1 enhances locomotor activity and c-fos expression following methamphetamine (MAP) challenge. (a) Horizontal locomotor activity was monitored before and after MAP (1.0 μg g^–1) or saline challenge at the adult stage. (b) Total locomotor activity was calculated and presented for the 60-min period after saline or MAP (1.0 or 2.0 μg g^–1) challenge (N=9–10 mice per group). (c) Vehicle-treated control and (d) NRG1β1-treated mice were subjected to c-fos immunohistochemistry 2 h after MAP (1.0 μg g^–1) challenge. Typical pictures of the medial prefrontal cortex (mpFC) are shown for mice challenged with MAP. Scale bar, 100 μm. (e) The number of c-fos-positive cells in the above microscopic field (725 × 965 μm) was counted bilaterally using the NIH Image cell-counting system (ver. 1.61) using five to seven sections of FC (+1.70 to +1.98 mm from the bregma), the nucleus accumbens (NAc) and striatum (+1.18 to +1.54 mm from the bregma), averaged for each mouse and subjected to statistical analysis (N=4 mice per group). Data are expressed as mean±s.e.m. ^*P<0.05, ^**P<0.01, compared between marked groups. ^#P<0.05, ^##P<0.01, ^###P<0.001, compared between MAP-challenged and -unchallenged groups by Fisher's least significant difference.

Full size image

We also examined c-fos expression in the brain following MAP challenge (1.0 μg g^–1) (Figure 5e). We found that MAP challenge induced a greater number of c-fos-positive cells in mpFC and the striatum in NRG1β1-treated mice than in control mice (both P<0.01) (Figures 5c–e). In contrast, there were no significant effects of NRG1β1 treatment in NAc (Figure 5e).

Neonatal NRG1β1 treatment impairs sensorimotor gating at the adult stage

We assessed sensorimotor gating of NRG1β1-treated mice by measuring PPI, which is often implicated in schizophrenia pathology or dopaminergic dysfunction.^{46, 47} We found a significant reduction of PPI in the NRG1β1-treated group (NRG1 treatment, F_(1, 22)=6.91, P<0.05) (Figure 6a) with no significant effect on pulse-alone startle (Figure 6b). As NRG1 signaling is involved in the survival of cochlear sensory neurons,⁴⁸ we also tested hearing ability by measuring auditory brainstem-evoked response thresholds. There were no differences in the auditory stimulus thresholds at any frequency between the two groups (Figure 6c).

Effects of neonatal exposure to neuregulin-1 (NRG1)β1 on hearing, startle responses and PPI. Neonatal mice were treated with NRG1β1 or vehicle (control) as described in Figure 2. At the adult stage, PPI levels, acoustic startle responses and brainstem-evoked response (ABR) thresholds were determined. (a) PPI was measured with 73, 76, 79 and 82 dB prepulse stimuli (N=12 mice per group). (b) Relative amplitudes of startle responses were monitored with 90, 95, 100, 105, 110, 115 and 120 dB tones (N=10 mice per group). (c) ABR thresholds were examined with specific auditory stimuli (4, 8, 10, 16, 20 and 32 kHz) by varying the sound pressure levels (N=9–10 mice per group). Pharmacological responses of NRG1β1-treated mice to risperidone were examined by PPI measurement. Mice daily received risperidone (1.0 μg g^–1, intraperitoneally) or saline during postnatal days (PNDs) 42–70. On 1 day after final administration of risperidone, PPI test was performed. (d) Effects of risperidone on pulse-alone startle (120 dB) were presented (N=10 mice per group). (e) Amelioration of PPI deficits in NRG1β1-treated mice by risperidone (N=10 mice per group). Data are expressed as mean±s.e.m. ^*P<0.05, ^**P<0.01, by Fisher's least significant difference.

Full size image

Risperidone, an atypical antipsychotic, has been shown to reduce psychotic symptoms and ameliorate PPI deficits in schizophrenia patients and in the animal models of schizophrenia.⁴⁹ Using a chronic administration schedule, we investigated the effect of risperidone (1.0 μg g^–1, daily) on PPI deficits in NRG1β1-treated mice. Risperidone administration significantly improved PPI deficits of NRG1β1-treated mice compared with control mice (drug, F_(1, 36)=7.46, P<0.01) (Figure 6e). However, it did not significantly alter the amplitudes of pulse-alone startle (Figure 6d).

Neonatal NRG1β1 treatment alters latent inhibition and social behaviors

We assessed the basal locomotor activity of adult mice, which influences behavioral evaluation of learning. There was no significant difference in locomotor activity scores between NRG1β1-treated and vehicle-treated control mice (NRG1 treatment, F_(1, 23)=1.17, P=0.29) (Supplementary Figure S7a). Then learning ability of these mice was measured with context- and tone-dependent fear conditioning. We found that NRG1β1 treatment in neonates did not alter learning performance in either paradigm (NRG1 treatment, F_(1, 20)=0.41, P=0.53 for the context-dependent learning; F_(1, 20)=0.21, P=0.66 for the tone-dependent learning) (Supplementary Figure S7b).

As NRG1β1-treated mice exhibited normal locomotion and fear learning, the mice were subjected to latent inhibition test. Latent inhibition of learning is considered to be an ability to ignore irrelevant stimuli and has been shown to involve the dopaminergic system.^{50, 51} This process is often disrupted in schizophrenia patients.⁵² Although basal learning scores were indistinguishable in non-pre-exposure (NPE) groups, the inhibitory effect of pre-exposure (PE) on freezing scores was different between vehicle-treated and NRG1β1-treated groups (NRG1 treatment × pre-exposure, F_(1,48)=4.23, P<0.05) (Figures 7a and b). In contrast to the significant latent inhibition of vehicle-treated mice (P<0.01), there was no significant difference in freezing rates between the NPE and PE groups of NRG1β1-treated mice (P=0.78) (Figure 7c), indicating their lack of latent inhibition.

Mice neonatally exposed to neuregulin-1 (NRG1)β1 show impaired latent inhibition and reduced social interaction. NRG1β1-treated and vehicle-treated mice were pre-exposed to tone cue (PE group) or not exposed to the cue (non-pre-exposure (NPE) group) and then subjected to tone-footshock pairs. On 1 day after conditioning, freezing rates (%) of (a) vehicle-treated control and (b) NRG1β1-treated mice were measured every 30 s in a different chamber before and during tone exposure. Insets reveal mean freezing rates during conditioning (N=12–14 mice per group). (c) Mean freezing rates during tone exposure in test trial. Social interaction was evaluated by the resident-intruder assay. We measured (d) time spent by the resident males over 10-min period actively pursing social investigation of the intruder mouse, (e) frequency of social behavior and (f) total time duration of fighting behavior (N=7–8 mice per group). Data are expressed as mean±s.e.m. ^*P<0.05, ^**P<0.01, by Fisher's least significant difference.

Full size image

To investigate the impact of neonatal NRG1β1 treatment on adult social behaviors in adulthood, we used a resident-intruder behavioral assay. In this assay, a group-housed male mouse (intruder) was placed in another home cage where a resident mouse had been housed alone until the test day. We found that male NRG1β1-treated mice (residents) showed a significant decrease in the duration and frequency of social interactions compared with vehicle-treated control male residents (both P<0.01) (Figures 7d and e). On the other hand, the duration of fighting behavior was indistinguishable between NRG1β1- and vehicle-treated control mice (Figure 7f).

In support of the neurodevelopmental hypothesis for the etiology of schizophrenia, here, we found the neuropathological implication of abnormal signals of the risk gene, NRG1. Specifically, we showed the profound influences of transient hyper-NRG1 signals on developing midbrain dopaminergic systems, evaluating the subacute and delayed neurochemical and behavioral consequences. Exogenous administration of NRG1β1 protein to mouse neonates produced the activation of midbrain ErbB4 receptors and caused marked changes in dopaminergic neurons and their associated behaviors: (1) Neonatal treatment of NRG1β1 increased the protein levels, phosphorylation, enzyme activity of TH and elevated dopamine levels in FC and/or the striatum; (2) In adulthood, NRG1β1-treated mice exhibited sustained increases in dopamine metabolism and depolarization-triggered dopamine release in FC; (3) NRG1β1-treated mice were more sensitive to MAP in regard to locomotor activity and c-fos induction; (4) NRG1β1-treated mice showed behavioral abnormalities in PPI, social interactions and latent inhibition. These results reveal a novel neurotrophic activity of NRG1 on developing midbrain dopaminergic neurons in vivo and indicate a biological link between prenatal or perinatal NRG1 and the dopaminergic pathology of schizophrenia. In contrast to the abnormalities in the dopaminergic systems described above, our examinations failed to uncover any gross structural deficits and neuronal or glial abnormalities with conventional dye staining and immunostatining. In addition, learning ability and basal locomotor activity were preserved in this animal model.

Influences of NRG1β1 on brain structure and function

Neuregulin-1 has a variety of neurotrophic activities on neuronal migration, axon guidance, myelination and synaptogenesis.^{12, 13, 14, 15} We assessed the influences of NRG1β1 on these neurotrophic processes in this model by quantifying the following neuronal and glial phenotypic markers in various brain regions: neuron-specific enolase, glial fibrillary acidic protein, myelin basic protein, 2′,3′-cyclic nucleotide 3′-phosphodiesterase, glutamate and GABA receptors (GluR1, GluR2/3, NMDAR1, NMDAR2A/B, GABAARα1), glutamate decarboxylase 65/67 and parvalbumin. However, there were no apparent neurochemical alterations in any of these molecular markers.

In contrast to these results, previous studies on knockout and transgenic mice of NRG1 or ErbB genes show that abnormal NRG1 signals result in phenotypic alterations such as impairments in myelin formation, migration of GABAergic interneurons and glial development.^{12, 13, 44, 45} We assume that the absence of phenotypic influences in our model could be due to (1) saturation of neurotrophic supports for these cells by other factors, (2) the distinct developmental time window of individual cell types or (3) difference in biological activities of individual NRG1 splicing variants.

Gene targeting in mice shows that the development of neuronal and glial cells are supported by multiple neurotrophic factors and cytokines.⁵³ Thus, signal redundancy by multiple neurotrophic factors can help explain the phenotypic differences seen between this NRG1β1 model and the genetic mutant models described earlier. Exogenously supplied NRG1 may not have pronounced effects on GABAergic neurons and glial cells if the same or similar factors endogenously provide a saturated level of neurotrophic support for these cells. Conversely, midbrain dopaminergic neurons might not receive enough neurotrophic support from endogenous NRG1 and be therefore competent to fully react to exogeneous NRG1 in this model. In this context, it is noteworthy that ErbB4 knockout mice did not exhibit any significant structural or neurochemical alterations in midbrain dopaminergic neurons.⁵⁴

In this study, we limited the exposure period of NRG1β1 to PNDs 2–10, which matches the developmental period of midbrain dopaminergic neurons.^{55, 56} In this context, it is possible that this cell population is relatively sensitive to NRG1β1 and dynamically reacts with this factor during the used experimental period. Thus, if we changed the timing of the NRG1β1 administration, the effect or affected cell population would differ.

Another possible explanation is based on the difference in biological activities of NRG1 splicing variants (types 1–5).^{57, 58, 59} Type-1 and type-2 NRG1 proteins comprise the extracellular Ig-like domain that interacts with heparan sulfate proteoglycans and modifies receptor binding and signaling.⁶⁰ In contrast, type-3 NRG1 has an activity to regulate oligodendrocyte development and myelination.¹³ The specific use of type-1 NRG1β1 in the present experiments might therefore limit its biological effects.

Neurotrophic activity of NRG1 on midbrain dopaminergic neurons

In this study, we found subacute effects of neonatal NRG1β1 treatment on developing dopaminergic systems. Interestingly, the influence of neonatal treatment with NRG1β1 was persistent in mpFC until adulthood. In adult mice treated with NRG1β1 as neonates, there were elevated dopaminergic innervation and metabolism. These results indicate that neonatal exposure to NRG1β1 can lead to lifelong impairment of dopamine synthesis and release in this brain region.

Dopaminergic innervation in the brain is classified into the three fiber routes: mesostriatal, mesolimbic and mesofrontocortical pathways. As ErbB4 mRNA is distributed in almost all classes of midbrain dopaminergic neurons in mice and primates,^{20, 21} these results raised a question regarding pathway specificity of the long-term effects of NRG1β1 treatment: Why was a dopaminergic abnormality persistent and apparent in mpFC? We would like to elaborate on the differences in developmental schedules among three dopaminergic pathways as follows.

In regard to the temporal schedule of development, there is significant time lag in the development of mesostriatal, mesolimbic and mesofrontocortical dopamine pathways. In rodents, the dopaminergic projections to the striatum and NAc are more extensive during midgestation.⁵⁵ In contrast, dopaminergic projections to the neocortical and limbic regions is vigorous at postnatal stages.^{56, 61} Thus, our hypothesis is that, at the postnatal stage when NRG1β1 was administered, corticolimbic fibers were still growing and capable of responding to this neurotrophic factor.

Behavioral similarity to genetic mutant mice

By targeting individual exons in the NRG1 gene, various types of knockout mice for NRG1 gene were developed, and their behavioral deficits were extensively investigated and compared.^{1, 16, 17, 19} There are some similarities in behavioral traits between NRG1β1-administered mice in this study and NRG1 knockout mice in published reports, even though these models presumably use an opposite strategy of NRG1 to reduce NRG1 signals. For example CRD-NRG1^+/− mice exhibited PPI deficits and Ig-NRG1^+/− mice have abnormalities in latent inhibition as seen in the present NRG1-injection model.^{16, 19} In this context, there is an interesting report that transgenic mice overexpressing type-1 NRG1 exhibited similar behavioral deficits in PPI.⁶²

Different types of NRG1 precursors, which are produced from a single gene, carry the distinct transmembrane domain(s) and are subjected to different modes of proteolytic regulation (for example, ectodomain shedding).⁵⁷ We established this animal model by stimulating ErbB receptors with processed type-1 NRG1 in a paracrine manner. This counterintuitive similarity between this model and these knockout mice might be explained by determining how individual exon disruption influences juxtacrine and paracrine/autocrine signaling of the remaining NRG1 variants and how it affects the molecular interaction of individual NRG1 variants with distinct ErbB receptors.^{12, 13, 15, 16, 63} Future studies should elucidate the biological characteristics of individual NRG1 variants and their precursors as well as their compensation or interference.^{57, 58, 59}

NRG1β1-treated mice as an animal model for schizophrenia

Schizophrenia patients show increased sensitivity to various dopamine agonists or releasers, such as amphetamine and cocaine.⁶⁴ Local or systemic administration of dopamine agonists to animals elicits the behavioral deficits in PPI as well as latent inhibition and social interactions that are implicated in schizophrenia,^{46, 51, 65} suggesting pathological contribution of hyperdopaminergic states to this disease.⁶⁶ In this study, neonatal NRG1β1 treatment resulted in similar behavioral deficits as well as persistent dopaminergic impairments in adulthood. In vivo microdialysis and monoamine measurement verified the hyperdopaminergic state of NRG1β1-treated mice. This argument is also supported by the hypersensitivity of this model to MAP. Furthermore, chronic treatment of risperidone ameliorated the deficits in PPI in NRG1β1-treated mice. Chronic antipsychotic treatment selectively decreases dopamine transmission in FC,⁶⁷ although different brain regions are also involved in PPI abnormality.⁶⁸ Although a plenty of controversies and discrepancies against the roles of dopamine in schizophrenia remain, our findings with the NRG1β1-treated mice provide evidence in favor of a hyperdopaminergic state in schizophrenia, at least within FC.^{68, 69, 70}

In contrast, the pathological link between the dopaminergic abnormality and the social deficits of this mouse model is controversial, as the reduction in social interaction is rather ascribed to the hypodopaminegic or serotonergic deficits.^{27, 71} In this context, we do not exclude the possibility that uncovered neurochemical deficits of NRG1β1-treated mice still remains to be explored.

Pathological implication of NRG1β1 in neurodevelopmental hypothesis of schizophrenia

The SNPs of the NRG1 gene that associate with the risk for schizophrenia are often located in the promoter region of NRG1 gene¹ and presumably involved in positive regulation of NRG1 gene transcription.⁵ In agreement, post-mortem studies support the association of increased NRG1 mRNA or protein with schizophrenia.^{3, 4, 5} Depending on the SNP type of NRG1, ischemic and traumatic brain injury may induce higher levels of NRG1 expression.^{10, 11} In this context, NRG1 is one of the candidate molecules that might be involved in both genetic and environmental vulnerabilities to schizophrenia.⁷²

Maternal infections and fetal/neonatal hypoxia are potential environmental risk factors for schizophrenia and are used in its animal modeling.^{7, 8, 73} Interestingly, the animal models established by the immune inflammatory insults also exhibit impaired dopaminergic innervations or metabolism.^{74, 75, 76} In this context, it is noteworthy that NRG1 is highly inducible in fetuses and neonates in response to these environmental insults.^{9, 10} Therefore, it is possible that NRG1 and potentially other ErbB4 ligands might contribute to the dopaminergic impairment of these animal models as well as that of schizophrenia.^{25, 26}

On the basis of the neurodevelopmental hypothesis,^{24, 25, 26} we have tested various inflammatory cytokines and neurotrophic factors to address the question of whether they can mediate the environmental insults for schizophrenia risk.⁷⁷Among many factors examined, neonatal treatment only with epidermal growth factor, interleukin-1 and NRG1 produce the long-lasting behavioral impairments that are implicated in schizophrenia models.^{40, 66, 78} Interestingly, these factors have a common neurotrophic activity on midbrain dopaminergic neurons.^{30, 40} However, we did not detect any neurobehavioral influences of control proteins (cytochrome c and albumin) in the in vivo experimental paradigm.^{40, 77, 78}

In conclusion, NRG1 is one of the key neurotrophic factors that have crucial impact on dopaminergic development and its neuropathology. We hope that this model established with NRG1 will facilitate the validation of both neurodevelopmental and dopaminergic hypotheses of schizophrenia.

HN was a recipient of the research grants from Astrazeneca Pharmaceutical Inc.

We are grateful to Dr Jourdi for proofreading. This work was supported by Health and Labor Sciences Research Grants, a grant for Promotion of Niigata University Research Projects, Core Research for Evolutional Science and Technology from the JST Corporation and a grant-in-aid from the Ministry of Health, Labor and Welfare, Japan.

Affiliations

1. Department of Molecular Neurobiology, Brain Research Institute, Niigata University, Niigata, Japan

   • T Kato
   • , Y Abe
   • , H Sotoyama
   •  & H Nawa

2. Department of Pathological Neuroscience, Resource Branch for Brain Disease Research, Brain Research Institute, Niigata University, Niigata, Japan

   • A Kakita
   •  & H Takahashi

3. Department of Molecular Genetics, Niigata University, Niigata, Japan

   • R Kominami
   •  & S Hirokawa

4. Consolidated Research Institute for Advanced Science and Medical Care, Waseda University, Tokyo, Japan

   • M Ozaki

Corresponding author

Correspondence to H Nawa.

To obtain permission to re-use content from this article visit RightsLink.

Publication history

DOI

https://doi.org/10.1038/mp.2010.10