Sustained brain-derived neurotrophic factor up-regulation and sensorimotor gating abnormality induced by postnatal exposure to phencyclidine: comparison with adult treatment

Authors


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

Abstract

Brain-derived neurotrophic factor (BDNF) is involved in synaptic development and plasticity, and alterations in BDNF expression or signaling are implicated in drug addiction and psychiatric diseases, such as depression and schizophrenia. In this study, we administered phencyclidine to postnatal and adult rats with different time schedules, and determined the correlations between BDNF expression and the behavioral effects. Both single and repeated phencyclidine injections into adult rats induced BDNF up-regulation in the corticolimbic system and a decrease in prepulse inhibition, both of which were transient. In contrast, subchronic postnatal administration increased BDNF protein and mRNA levels in the hippocampus and entorhinal cortex, which were sustained until 8 weeks of age. In parallel, the postnatal rats treated with phencyclidine developed a persistent decrease in prepulse inhibition at the adult stage. The chronic BDNF increase appeared to contribute to the prepulse inhibition abnormality, as subchronic BDNF infusion into the hippocampus of normal rats mimicked the prepulse inhibition deficits. This study suggests that phencyclidine exposure during brain development induces sustained BDNF up-regulation in the limbic system with a biological link to sensorimotor gating deficits.

Abbreviations used
BDNF

brain-derived neurotrophic factor

EIA

enzyme immunoassay

PCP

phencyclidine

PND

postnatal day

PPI

prepulse inhibition

SDS

sodium dodecyl sulfate

SSC

saline-sodium citrate

Brain-derived neurotrophic factor (BDNF) is regulated via glutamate receptors in an activity-dependent manner (Zafra et al. 1990; Patterson et al. 1992; Castren et al. 1993). BDNF also exerts strong trophic activity to support the survival of midbrain dopaminergic neurons, triggers dopamine release and regulates dopamine receptor expression (Hyman et al. 1991; Zhou et al. 1996; Guillin et al. 2001). Thus, BDNF may mediate the glutamatergic modulation of the dopaminergic system. Indeed, dysregulation of BDNF production or release is associated with neuropsychiatric disorders, such as drug addiction and schizophrenia, both of which involve dopaminergic and glutamatergic dysfunction (Kelley 2004; Harrison and Weinberger 2005). Amphetamine-induced stereotypical behavior and cocaine craving are accompanied by BDNF up-regulation in the corticolimbic areas (Meredith et al. 2002; Grimm et al. 2003). In patients with chronic schizophrenia, BDNF protein levels are increased in the hippocampus and anterior cingulate cortex and, conversely, reduced in the prefrontal cortex (Takahashi et al. 2000; Weickert et al. 2003).

The involvement of the dopaminergic and glutamatergic systems in neuropsychiatric disorders has been studied extensively using phencyclidine (PCP) (Jentsch and Roth 1999), which inhibits NMDA receptor activity. PCP administration in humans causes a psychotomimetic state that closely resembles schizophrenia (Javitt and Zukin 1991). In adult rodents, PCP induces behavioral changes analogous to those in psychotic patients: hyperlocomotion, learning impairment, decreased social interaction and deficits in prepulse inhibition (PPI) of the acoustic startle response (Sams-Dodd 1998; Jentsch and Roth 1999). Although most of the behavioral changes are preserved even after subchronic treatment with PCP (Sams-Dodd 1998; Jentsch and Roth 1999), PPI deficits are observed only during treatment and not after its cessation (Martinez et al. 1999). In contrast, changes after PCP exposure in the early postnatal period are more long-lasting. Hyperlocomotion and learning impairment are sustained until adulthood (Brooks et al. 1997b; Wang et al. 2001). Moreover, PPI deficits are observed during development [postnatal day (PND) 24–26] (Wang et al. 2001). There are no reports, however, of neurochemical data that closely illustrate the temporal pattern of PCP-induced behavior, especially PPI deficits.

Linden et al. (2000) reported that acute injection of a non-competitive NMDA receptor antagonist, MK-801, increased BDNF mRNA and protein levels in the entorhinal cortex. In this study, we treated both postnatal and adult rats with PCP and compared the effects of PCP on brain BDNF levels and PPI. In parallel, BDNF levels were also determined in mutant mice deficient for the NMDA receptor subunit gene(s) and compared with those induced by PCP. Finally, we tested whether chronic brain infusion of BDNF induces PPI deficits, and discuss the possibility that abnormal BDNF signaling impairs PPI.

Materials and methods

Animals

Male 8-week-old Wistar rats and newborn Wistar and maternal rats were purchased from Japan SLC, Inc. (Shizuoka, Japan). NMDA receptor GluRε1 (NR2A) and GluRε2 (NR2B) mutant mice were generated using TT2 embryonic stem cells derived from C57BL/6 × CBA F1 hybrid mice, as described previously (Sakimura et al. 1995; Kutsuwada et al. 1996). Male homozygous mutant mice lacking GluRε1 (GluRε1–/–) and heterozygous mutant mice deficient in GluRε2 (GluRε2+/–) were used. Rat pups were housed, nine to ten per litter, with their mother, and, after weaning on PND21, were housed three to four per cage. Adult rats and mice were housed three to four per cage. All animals were maintained on a 12-h light–dark cycle with free access to food and water. All of the animal experiments described were performed in accordance with the institutional guidelines on the ethical use of laboratory animals.

Drug administration

For adult PCP exposure, male 8-week-old rats received a single subcutaneous injection of PCP (10 mg/kg, Mitsubishi Welpharma, Saitama, Japan) or saline as an acute experiment. In the subchronic experiment, rats were injected subcutaneously with the same dose of PCP for 14 consecutive days. For early postnatal PCP exposure, 2 or 10 mg/kg PCP, dissolved in physiological saline, was injected subcutaneously into rat pups for 14 days from PND 3 to PND 16. Littermates injected with saline were used as controls for all experiments. The body weight of PCP-treated rats averaged 23% less than that of saline-treated rats at PND 50. PCP-treated rats appeared to be healthy, however, and the weight gain was comparable with that of saline-treated rats after weaning. Postnatal PCP treatment had no significant effect on the weight of the brain tissues dissected (see below).

BDNF enzyme immunoassay (EIA)

Rats were killed with CO2 and decapitated, either 12 or 24 h after administration of a single dose of PCP or 1, 7, 14 or 28 days after the end of subchronic administration. Brains were rapidly removed and tissues were identified according to a rat brain atlas (Paxinos and Watson 1986). Prefrontal cortex, parietal cortex, anterior and posterior cingulate cortex, hippocampus and entorhinal cortex were taken, immediately frozen in dry ice, and stored at − 80°C until use. BDNF levels were measured by two-site EIA, as described previously (Nawa et al. 1995). Briefly, anti-BDNF antisera were prepared and used as the primary antibody in BDNF detection to coat polystyrene 96-well microtiter plates for EIA. The high-affinity anti-BDNF antibodies were biotinylated with NHS-LC-Biotin and used as the secondary antibody. The EIA detected trace amounts of BDNF (1 pg/assay) and did not cross-react with 1000-fold excess amounts of other neurotrophins (Nawa et al. 1995).

Northern blot analysis

Total RNA was extracted from brain tissues by ISOGEN (Nippon Gene, Tokyo, Japan). RNA samples (24 µg) were denatured in 50% formamide, 6% formaldehyde, 20 mm3-[N-morpholino]propanesulfonic acid (MOPS) buffer (pH 7.0) and 1 mm EDTA at 65°C for 10 min, and fractionated on a formaldehyde-agarose gel. The gels were treated with water for 30 min, and the RNA was transferred to a Hybond-N+ nylon membrane (Amersham Pharmacia Biotech, Piscataway, NJ, USA) with 20 × saline-sodium citrate (SSC) buffer and fixed by exposure to ultraviolet light. 32P-labeled cDNA probes were synthesized from template BDNF or β-actin cDNA using the hexanucleotide random priming method (Feinberg and Vogelstein 1983). Hybridization was performed at 42°C for 12–18 h in a solution containing 50% formamide, 5 × SSC, 5 × Denhardt's solution and 1% sodium dodecyl sulfate (SDS). The membranes were washed with 0.3 × SSC, 1% SDS at 65°C. The radioactivity of positive RNA bands was directly measured with background compensation using a Fuji Bioimage Analyzer (BAS5000; Fuji Medical System, Tokyo, Japan). BDNF mRNA levels were normalized with β-actin mRNA levels for statistical comparison.

Western blot analysis

Relative levels of GluR subunits were determined by western blot analysis. Brain tissues were homogenized with tissue lysis buffer containing 200 mm Tris and 2% SDS. After centrifugation at 10 000 g for 10 min, protein in supernatants was boiled and protein concentrations were determined using a Micro BCA kit (Pierce Chemical, Rockland, IL, USA). Protein samples (30 µg/lane) were denatured in Laemmli sample loading buffer (10% glycerol, 2% SDS, 20 mm dithiothreitol and 63 mm Tris) at 90°C, separated by SDS-polyacrylamide gel electrophoresis, and transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA, USA) by electrophoresis. The membrane was probed with anti-NR1 antibody (1 : 500), anti-NR2A/B antibody (1 : 200; Chemicon, Temucula, CA, USA), anti-GluR1 antibody (1 : 1000), anti-GluR2/3 antibody (1 : 1000) and anti-β-actin antibody (1 : 1000; Chemicon) at 4°C overnight. Anti-NR1, anti-GluR1 and anti-GluR2/3 antibodies were produced in our laboratory (Ibaraki et al. 1999; Narisawa-Saito et al. 1999). After extensive washing, the immunoreactivity on the membrane was detected using an anti-rabbit immunoglobulin conjugated to horseradish peroxidase (1 : 10 000), followed by a chemiluminescence reaction (ECL kit; Amersham Pharmacia Biotech). The immunoreactivity of the bands was quantified by densitometric analysis. Levels of NR1, NR2A/B, GluR1 and GluR2/3 were normalized with β-actin levels.

Histology

Rats were deeply anesthetized by inhalation of diethyl ether, and killed by transcardial perfusion with 4% paraformaldehyde plus 0.1% glutaraldehyde in 0.1 m phosphate buffer solution (pH 7.4). The brains were removed, immersed in the same fixative for 24 h, dehydrated in a graded ethanol series, and embedded in paraffin wax. Serial sections, 4 µm thick, were cut from the blocked samples and stained with either hematoxylin and eosin or Klüver-Barrera stain.

Measurement of acoustic startle and PPI of startle

Acoustic startle and PPI responses were measured in a startle chamber (SR-Laboratory Systems, San Diego Instruments, San Diego, CA, USA) adapted for rats. The paradigm was adapted from Bakshi and Geyer (1998) and used to assess startle amplitude and PPI response with acoustic stimuli of 120 dB, a single pre-pulse interval (100 ms) and three different pre-pulse intensities [5, 10 and 15 dB above background noise (white noise, 70 dB)]. Each rat was placed in the startle chamber and initially acclimatized for 5 min with background noise alone. The rat was then subjected to 40 startle trials, each trial consisting of one of five conditions: (i) a 40-ms, 120-dB noise burst presented alone; (ii–iv) a pre-pulse (20-ms noise burst) that was 5, 10 or 15 dB above background noise (i.e. 75, 80 or 85 dB) followed, 100 ms later, by a 40-ms, 120-dB noise burst; (v) no stimulus (background noise alone), which was used to measure baseline movement in the chamber. These five trial types (i–v) were each repeated eight times in a pseudorandom order to give 40 trials. The intertrial interval was 15 s. Each trial type was presented once within a block of five trials. The percentage PPI of a startle response was calculated as: 100 − [(startle response on pre-pulse − pulse stimulus trials − no stimulus trials)/(pulse-alone trials − no stimulus trials] × 100).

Single or continuous BDNF infusion into the hippocampus

To evaluate BDNF effects on PPI, BDNF was administered acutely or continuously to the hippocampus. Surgery was performed on male 8-week-old Wistar rats as described previously (Croll et al. 1994). Briefly, a guide cannula or a cannula was implanted unilaterally into the right hippocampus (4.3 mm posterior and 2.5 mm lateral measured from the bregma). Rats were allowed to recover from surgery for more than 3 days, and 4.0 µg of recombinant human BDNF (n = 8, 1 µg/µL; Sumitomo Pharmaceuticals, Osaka, Japan), or the same volume of saline (n = 8), was infused over 30 min from a Hamilton syringe. Alternatively, using Alzet osmotic minipumps (model 2002; Cupertino, CA, USA), saline (n = 5) or BDNF (n = 5, 1 µg/µL) was infused continuously. PPI was measured 30 min after acute infusion or after 10 days of continuous infusion.

Statistical analysis

All values are presented as the mean ± SEM. BDNF values and neurochemical data were analyzed using the unpaired t-test. PPI data were subjected to one-way analysis of variance (anova) with repeated measures. In the evaluation of postnatal PCP effects, the initial anova failed to prove sex and brain region differences. Accordingly, BDNF values were analyzed separately for each brain region by one-way anova with post hoc Dunnett's test. A probability level of p < 0.05 was considered to be statistically significant.

Results

Transient BDNF up-regulation by PCP in adult rats

Adult rats were treated with PCP to determine whether PCP induces the same BDNF increase as observed with MK-801 (Linden et al. 2000). Twelve hours after a single injection of 10 mg/kg PCP, there was a significant increase in BDNF levels in the prefrontal cortex (58.3% increase, p < 0.01), parietal cortex (140.5% increase, p < 0.001), anterior cingulate cortex (56.5% increase, p < 0.00), posterior cingulate cortex (158.0% increase, p < 0.001) and entorhinal cortex (56.6% increase, p < 0.05) (Fig. 1a), consistent with Linden et al. (2000). In the anterior cingulate cortex and posterior cingulate cortex, the increase remained significant for 24 h after the injection (anterior cingulate cortex, 38.4% increase, p < 0.05; posterior cingulate cortex, 45.9% increase, p < 0.01; Figs 1b and c). There was no significant increase in any other brain region 24 h after the injection. We also measured BDNF levels in the brains of rats injected daily with 10 mg/kg PCP for 14 days. Although there was a significant increase in BDNF levels in the posterior cingulate cortex 24 h after the last injection (64.8% increase, p < 0.05; Fig. 1c), the increase disappeared within 7 days.

Figure 1.

 Temporal pattern of brain-derived neurotrophic factor (BDNF) level changes in adult rats after acute and subchronic phencyclidine (PCP) exposure. BDNF was extracted for duplicate enzyme immunoassay (EIA) determinations from the prefrontal cortex (PFC), parietal cortex (PRC), anterior cingulate cortex (ACC), posterior cingulate cortex (PCC), cingulate cortex (CGC), hippocampus (HIP) and entorhinal cortex (ENR). (a) Adult rats (8 weeks) were killed 12 h after a single injection of 10 mg/kg PCP (n = 9) or saline (SAL, n = 8). The bar graph shows the mean with SEM values. (b, c) Time course of BDNF levels in ACC (b) and PCC (c) after acute or subchronic PCP administration. Rats were killed 12 or 24 h after a single injection or 1, 7, 14 or 28 days after repeated injections of 10 mg/kg PCP or SAL. Each plot represents the mean with SEM values of six to eight samples. In PFC, PRC and ENR, BDNF levels were comparable between PCP and SAL groups 24 h after a single injection. *p < 0.05, **p < 0.01, ***p < 0.001 compared with SAL by unpaired t-test.

Sustained BDNF overexpression in rats postnatally exposed to PCP

To examine whether PCP affects BDNF expression during brain development, rat pups were treated daily with 10 mg/kg PCP or saline for 14 days (PND 3–16) and BDNF levels were measured 24 h after the last injection (PND 17). There were no significant differences in the different sexes (F1.75 = 0.2, p = 0.63), and therefore the male and female data were combined. BDNF protein levels were increased in the hippocampus (26% increase, p < 0.05) and entorhinal cortex (31% increase, p < 0.05) of postnatal rats treated daily with PCP (PND 17), compared with saline-treated controls (Fig. 2). There was also a non-significant increase in BDNF levels in the prefrontal cortex (17.0%) after PCP treatment. BDNF levels were not changed in the other brain regions examined.

Figure 2.

 Increased limbic brain-derived neurotrophic factor (BDNF) levels in postnatal rats treated daily with phencyclidine (PCP). Rat pups were treated daily with 10 mg/kg PCP (male/female = 5/4) or saline (SAL, male/female = 4/6) for 14 days [postnatal day (PND) 3–16], and killed 24 h after the last injection (PND 17). BDNF was extracted from the prefrontal cortex (PFC), parietal cortex (PRC), cingulate cortex (CGC), hippocampus (HIP) and entorhinal cortex (ENR), and the levels were determined by enzyme immunoassay (EIA). The bar graph shows the mean with SEM values obtained from 9 to 10 samples. *p < 0.05 compared with SAL by unpaired t-test.

Postnatal rats were similarly treated with PCP, but BDNF levels were measured at the adult stage. There were significant interactions between brain region and drug treatment (F10.120 = 2.1, p < 0.05). The BDNF values were analyzed separately in each brain region after combining the male and female data, because there was no effect of sex (F1.120 = 0.1, p = 0.75). The BDNF increase was observed in rats up to 8 weeks of age in the hippocampus (25% increase; F2.23 = 8.5, p < 0.01) and entorhinal cortex (30% increase; F2.23 = 5.0, p < 0.05; Fig. 3a). In the prefrontal cortex and other brain regions, BDNF levels were comparable between PCP- and saline-treated rats (data not shown). Northern blot analysis revealed that the increase in hippocampal BDNF protein was accompanied by an increase in BDNF mRNA levels (67.1% increase, p ≤ 0.01; Fig. 3b). These results indicate that BDNF protein and mRNA levels are up-regulated in the limbic system in postnatal rats challenged daily with PCP, and, in contrast with adult PCP treatment, the up-regulation is sustained until adulthood.

Figure 3.

 Sustained brain-derived neurotrophic factor (BDNF) overexpression in the limbic system after postnatal exposure to phencyclidine (PCP). (a) Rat pups were treated daily with 2 mg/kg PCP (male/female = 7/2), 10 mg/kg PCP (male/female = 5/3) or saline (SAL, male/female = 5/4) for 14 days [postnatal day (PND) 3–16], and killed at 8 weeks of age. The bar graph shows the mean with SEM values obtained from 8 to 10 samples. One-way anova revealed a significant drug effect [hippocampus (HIP), F2.23 = 8.5, p < 0.01; entorhinal cortex (ENR), F2.23 = 5.0, p < 0.05]. *p < 0.05, **p < 0.01 compared with SAL by post hoc Dunnett's test. (b) Northern blotting analysis of BDNF mRNA expression in the hippocampus. Total RNA was extracted from brain tissues obtained from 8-week-old rats postnatally exposed to 10 mg/kg PCP (n = 4) or SAL (n = 4). A northern blot was first probed with BDNF cDNA and reprobed with β-actin cDNA to normalize BDNF mRNA signals. The bar graph shows the relative amounts of the sum of both BDNF mRNA splice variants indicated as arrows. The sizes of the variants match those reported previously (Maisonpierre et al. 1991). **p < 0.01 compared with SAL by unpaired t-test. The original radioactivity for BDNF mRNA was weak and measured directly using a Fuji Bioimage Analyzer (BAS5000). A digital image of the autoradiograph is enhanced for display.

BDNF down-regulation in NMDA receptor subunit mutant mice

If sustained NMDA receptor blockade by PCP leads to BDNF up-regulation, a developmental lack or decrease in NMDA receptor subunit expression and function might also result in a BDNF increase. We therefore measured brain BDNF levels in adult mice (90 days of age) lacking the GluRε1 gene (GluRε1–/–) or deficient in the GluRε2 gene (GluRε2+/–). BDNF protein levels tended to be decreased in both GluRε1–/– and GluRε2+/– mice, compared with those of control littermates (Fig. 4). The BDNF decrease in GluRε1–/– mice was not significant. There was a significant decrease in BDNF levels in GluRε2+/– mice in the prefrontal cortex (42.3% decrease, p < 0.01), parietal cortex (39.9% decrease, p < 0.05), cingulate cortex (55.1% decrease, p < 0.001), hippocampus (35.3% decrease, p < 0.05) and entorhinal cortex (52.7% decrease, p < 0.01). These results suggest that early postnatal effects of daily PCP exposure on BDNF expression cannot be ascribed directly to sustained NMDA receptor blockade.

Figure 4.

 Brain-derived neurotrophic factor (BDNF) levels in the brain in NMDA receptor knockout mice. Brain tissues [prefrontal cortex (PFC), parietal cortex (PRC), cingulate cortex (CGC), hippocampus (HIP) and entorhinal cortex (ENR)] were obtained from 90-day-old GluRε1–/– (epsilon 1, n = 6) and GluRε2+/– (epsilon 2, n = 6) mutants, and wild-type mice (n = 6). The bar graph shows the mean with SEM values obtained from duplicate enzyme immunoassay (EIA) determinations. *p < 0.05, **p < 0.01, ***p < 0.001 compared with wild-type mice by unpaired t-test.

Induction of NR1 subunits by postnatal PCP exposure

To explore the difference in BDNF expression between PCP-treated rats and NMDA receptor knockout mice, relative protein levels of glutamate receptor subunits were measured in the hippocampus of rats postnatally treated with PCP. Postnatal PCP treatment significantly increased the NR1 NMDA receptor subunit (27.2%, p ≤ 0.05) at the adult stage (Fig. 5). The GluR2/3 α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and NR2A/B NMDA receptor subunit levels also tended to be increased (GluR2/3, 34.6% increase; NR2A/B, 15.6% increase), although the difference was not significant.

Figure 5.

 Relative protein levels of glutamate receptor (GluR) subunits in the hippocampus of rats postnatally exposed to phencyclidine (PCP). Protein was extracted from the brain tissues obtained from adult rats (8 weeks) postnatally exposed to 10 mg/kg PCP (n = 4) or saline (SAL, n = 4), and was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, transferred to a polyvinylidene difluoride membrane, and probed with antibodies directed against NR1, NR2A/B, GluR1, GluR2/3 or β-actin. Immunoreactive bands for β-actin are shown as loading controls. Results of densitometric measurements are normalized to β-actin levels and shown as means with SEM. *p < 0.05 compared with SAL by unpaired t-test.

Brain histology of rats postnatally exposed to PCP

A conventional histological study at PND 17 revealed no drastic reduction in neuronal cell number and no signs of neuronal damage, such as eosinophilic neurons or microglia in the cortical areas and hippocampus of rats, after postnatal treatment with 10 mg/kg PCP (Fig. 6). There was no sign of chromatin condensation. Similar negative findings were also observed in 8-week-old rats (data not shown), suggesting a lack of neurodegeneration. The lack of neuronal loss in the hippocampus is supported by a recent study that examined subchronic effects of PCP on neurodegeneration (Wang and Johnson 2005).

Figure 6.

 Histological examination of brains postnatally exposed to phencyclidine (PCP). Brain sections through the parietal cortex (a, d), hippocampus (b, e) and cerebellum (c, f) of control (a–c) and postnatally PCP-treated (d–f) rats were examined at postnatal day (PND) 17 with the Klüver-Barrera stain. Scale bars, 0.3 mm. Hematoxylin and eosin staining of the CA1 pyramidal layer or Purkinje layer is shown in the subpanels. Repeated staining revealed no apparent differences in adjacent sections over three animals.

Effects of postnatal and adult PCP treatments on PPI

PPI was measured in 8-week-old rats that had been repeatedly treated with 2 mg/kg PCP, 10 mg/kg PCP or saline during the postnatal stage. PPI values of males and females were combined because the initial anova revealed no significant differences between sexes (F1.63 = 1.0, p = 0.33). Postnatal PCP exposure had a significant effect on PPI at the adult stage (F2.24 = 4.6, p < 0.05; Fig. 7a). Post hoc tests revealed that postnatal rats exposed to daily PCP challenges had significantly lower PPI to some pre-pulse intensities (Fig. 7b).

Figure 7.

 Abnormal sensorimotor gating induced by treatment with phencyclidine (PCP). Sensorimotor gating was examined in rats (8 weeks) postnatally exposed to 2 mg/kg PCP (male/female = 7/2), 10 mg/kg PCP (male/female = 5/4) or saline (SAL, male/female = 6/3) (a). The bar graph shows the mean with SEM of prepulse inhibition (PPI) to three prepulse intensities. anova with repeated measures revealed a significant drug effect on PPI (F2.24 = 4.6, p < 0.05). PPI was measured 2 or 24 h after a single PCP injection in adult rats (2 h, b; 24 h, c) or 24 h after the end of subchronic PCP administration (d). Each plot represents the mean with SEM values of eight to nine samples. anova with repeated measures revealed a significant drug effect (F1.15 = 24.6, p < 0.001) on PPI measured 2 h after a single PCP injection (b). *p < 0.05, **p < 0.01, ***p < 0.001 compared with SAL by post hoc Dunnett's test.

We also monitored the PPI levels of rats that were acutely or repeatedly challenged to PCP at the adult stage. PPI deficits were robust in adult rats 2 h after a single PCP injection (F1.15 = 24.6, p < 0.001; Fig. 7b). There was a lack of PPI changes 24 h after a single injection or repeated daily injections of PCP (Figs 7c and d), paralleling the transient BDNF up-regulation after adult PCP treatment.

PPI deficits after hippocampal BDNF infusion

To evaluate the neurobehavioral impact of BDNF up-regulation on PPI, BDNF was acutely or continuously administered into the hippocampus of naive adult rats with an implanted cannula or an osmotic minipump. No seizures or abnormal behaviors were observed in any of the rats throughout the hippocampal BDNF infusion period (data not shown). PPI was measured 30 min and 10 days after acute BDNF injection and pump implantation, respectively, and compared between BDNF- and saline-infused animals. The acute single infusion of BDNF had no effect on PPI (Fig. 8a). In contrast, chronic BDNF administration significantly decreased PPI (F1.8 = 6.0, p < 0.05; Fig. 8b). This result suggests an involvement of sustained BDNF signals in the PPI disruption observed after postnatal PCP exposure.

Figure 8.

 Pre-pulse inhibition (PPI) changes induced by acute (a) or subchronic (b) brain-derived neurotrophic factor (BDNF) infusion into the hippocampus. Surgery was performed on 8-week-old male Wistar rats. (a) BDNF (n = 8) or saline (SAL, n = 8) was injected to the hippocampus through the implanted guide cannula. Thirty minutes after injection, PPI was meausured. (b) BDNF (n = 5) or saline (SAL, n = 5) was infused into the hippocampus for 10 days and PPI was measured. *p < 0.05 by one-way anova with repeated measures (F1.8 = 6.0).

Discussion

The data presented here demonstrate that PCP treatment up-regulates BDNF in both adult and postnatal rats. In contrast with the transient effects of PCP in the adult stage, postnatal PCP treatment triggers a BDNF increase that is sustained until adulthood. Postnatal PCP treatment induces a persistent sensorimotor gating disturbance at adulthood, whereas PCP treatment in adults results in only a transient PPI deficit. The main finding of this study is that exposure of the developing brain to PCP triggers a sustained BDNF up-regulation associated with a long-lasting neurobehavioral impairment.

Regulation of BDNF by NMDA receptor activity

The results presented here replicate earlier findings indicating that acute MK-801 injection increases BDNF levels in adult rat brain (Linden et al. 2000), and further extend this finding by demonstrating sustained BDNF up-regulation after postnatal PCP exposure. GluRε1 (NR2A) and GluRε2 (NR2B) subunits are sensitive to the non-competitive antagonists, MK-801 and PCP (Yamakura et al. 1993), and have a central role in hippocampal long-term potentiation (Sakimura et al. 1995; Rostas et al. 1996). In contrast with the pharmacological effects of daily PCP treatment, the present data on GluRε1 (NR2A) and GluRε2 (NR2B) mutant mice demonstrate that life-long or more widespread NMDA receptor hypofunction down-regulates BDNF expression. This is consistent with the fact that BDNF is regulated in an excitatory neurotransmission-dependent manner (Zafra et al. 1990; Patterson et al. 1992; Castren et al. 1993). NMDA receptor hyperactivity, triggered by subchronic PCP treatment, is known to be attenuated by anti-psychotic drugs (Ninan et al. 2003). Chronic treatment with haloperidol, risperidone or olanzapine decreases BDNF levels in several brain regions (Angelucci et al. 2000, 2005). Furthermore, pre-treatment with haloperidol or clozapine attenuates the MK-801-induced increase in BDNF mRNA levels (Linden et al. 2000), suggesting that anti-psychotic drugs may have a normalizing effect on PCP-induced BDNF increases.

Earlier studies have reported that postnatal PCP increases NR1 subunit levels in the frontal cortex and hippocampus (Wang et al. 2001; Harris et al. 2003; Sircar 2003). Our western blot data replicated these reports. Although it is not clear how postnatal PCP treatment up-regulates GluR subunits, NMDA receptor blockade has been reported to induce compensatory glutamate release that can stimulate non-NMDA glutamate receptors, which are not antagonized by PCP (Moghaddam and Adams 1998). The observed difference in the genetic and pharmacological effects of NMDA receptor blockade may strengthen the argument that partial blockade of the NMDA receptor function, which manifests as a psychotomimetic state during withdrawal from PCP, may have a central role in BDNF up-regulation (see below) (Holcomb et al. 2005).

This study has revealed a substantial difference in the temporal and spatial patterns of PCP effects between postnatal and adult brain. Postnatal daily PCP exposure increases BDNF levels in the limbic system, which are sustained until adulthood. In contrast, PCP effects on BDNF in adults are absent in the hippocampus, transient in the cingulate cortex and develop tolerance to subchronic treatment. The pharmacological effects of repeated PCP challenges are consistent with the effects of reversible neurotoxic cytoplasmic vacuolization, which is induced by single PCP injection, mainly in the posterior cingulate cortex, but subsides within 24 h, and develops tolerance to repeated injections (Olney and Farber 1995). Activation of NMDA receptors on interneurons maintains inhibitory control on excitatory neurons in the posterior cingulate cortex and other cortical areas (Olney and Farber 1995). MK-801-induced disinhibition has a central physiological function in these interneurons (Li et al. 2002), and subsequent neuronal excitation is regarded as a cause of cytoplasmic vacuolization. The same mechanism may apply to the adult PCP-induced BDNF up-regulation that is predominant in the posterior cingulate cortex.

It is not clear why the postnatal limbic system is so sensitive to PCP. Synapse formation in the hippocampus is the most immature in early postnatal rodent brain (Super et al. 1998), and correct neuronal activity is important for the development of hippocampal circuitry (Lauri et al. 2003). In hippocampal slice cultures, chronic blockade of NMDA receptors in the postnatal period leads to an increased number of excitatory synaptic connections (Luthi et al. 2001). On the other hand, BDNF is involved in synapse formation, as demonstrated by its influence on the development of ocular dominance columns (Gianfranceschi et al. 2003) or whisker representation in the barrel cortex (Genoud et al. 2004). In this study, there was a decrease in body weight during and after postnatal PCP treatment, suggesting that the treatment potentially produces nutritional deficits for brain development as well. Brooks et al. (1997a) reported a decrease in synaptic density in the occipital cortex, which goes through a rebound increase afterwards. These studies suggest that postnatal PCP challenge perturbs synaptogenesis in the critical period of hippocampal development, potentially through the blockade of NMDA receptors and/or BDNF up-regulation.

Negative effect of hippocampal BDNF on sensorimotor gating

PPI of the acoustic startle response is a measure of sensorimotor gating (Geyer et al. 2001), which is applicable to humans as well as rats. PPI impairments have been reported in patients with schizophrenia and other neuropsychiatric disorders (Braff et al. 2001; Geyer et al. 2001), thus providing a neurobiological basis common to humans and animals. Several studies have demonstrated the involvement of the dopaminergic and glutamatergic systems in sensorimotor gating (Bakshi and Geyer 1998; Geyer et al. 2001; Takeuchi et al. 2001). NMDA receptor blockade consistently disrupts PPI in the acute phase (Geyer et al. 2001). Postnatal daily PCP exposure induces persistent PPI deficits during development (Wang et al. 2001), whereas repeated PCP administration fails to induce sustained PPI disruption in adult rats (Ehrhardt et al. 1999; Martinez et al. 1999), both of which are consistent with the present results. This study has revealed a close association of BDNF up-regulation with this temporal pattern of PPI deficits. Interestingly, sensorimotor gating is rather enhanced in GluRε2 mutant mice (Kutsuwada et al. 1996; Takeuchi et al. 2001), in which hippocampal BDNF levels are decreased. Together with the present finding of BDNF down-regulation in this mutant, these previous observations indicate concomitant regulation of PPI and BNDF. The hippocampus is one of several limbic regions that mediate the PPI-disrupting effects of NMDA receptor antagonists (Bakshi and Geyer 1998). The present finding that continuous BDNF infusion into the hippocampus similarly disrupts PPI confirms that PCP-induced PPI deficits, at least to some extent, involve the up-regulation of BDNF. It has been shown that continuous BDNF infusion into the hippocampus causes mossy fiber sprouting and facilitates seizure activity (Scharfman et al. 2002). Limbic epileptogenesis has been reported to disrupt sensorimotor gating (Koch and Ebert 1998). These findings suggest that hippocampal hyperexcitability, which is presumably induced by continuous BDNF infusion and by the postnatal PCP-triggered BDNF up-regulation, contributes to PPI deficits.

Postnatal PCP exposure and synaptic plasticity

The postnatal limbic system is also sensitive to other manipulations. Maternal deprivation during the early postnatal period causes a persistent increase in hippocampal BDNF levels at adulthood (Kuma et al. 2004; Greisen et al. 2005), whereas other groups have reported that it produces a transient BDNF increase (Roceri et al. 2004). These observations indicate substantial similarity in BDNF expression between the maternal deprivation paradigm and postnatal PCP exposure, suggesting the implication of hippocampal BDNF in neurobehavioral development. In contrast, it has been reported that prenatal stress causes a decrease in NR1 and NR2B subunits without BDNF changes in the hippocampus (Fumagalli et al. 2004; Son et al. 2006), suggesting a substantial difference in interactions between BDNF and NMDA receptors.

In adult rats, chronic mild stress induces hyperphosphorylation of p42 and p44 mitogen-activated protein kinase (MAPK) in the prefrontal cortex (Trentani et al. 2002). Our preliminary data have indicated that postnatal PCP treatment significantly increases the phosphorylation of MAPK and decreases the phosphorylation of α-calmodulin-dependent kinase II (M. Takahashi, unpublished data). Further elucidation of changes in BDNF signaling may aid in the understanding of the aberrant synaptic plasticity or development that is induced by postnatal treatment with PCP.

Conclusion

Limbic BDNF up-regulation, induced by PCP exposure during a limited postnatal period, contributes to long-lasting neurobehavioral impairments. These findings may be implicated in the clinical observation of increased BDNF levels in the corticolimbic system of patients with chronic schizophrenia and in the temporal lobe of patients with epilepsy (Takahashi et al. 1999, 2000). The close association of BDNF levels with sensorimotor gating suggests that neurotrophic conditions severely influence the later development of sensorimotor gating. This study reveals a novel biological link between postnatal neurotrophic states and the functional development of sensorimotor gating.

Acknowledgements

We thank K. Araki, S. Hashimoto and E. Higuchi for technical assistance, and Dr M. Mishina for authorizing the use of knockout mice. We are also grateful to Sumitomo Pharmaceuticals for recombinant BDNF. This work was supported by the Targeted Research Grant for Brain Sciences, Grant-in-Aid for Creative Scientific Research and Grants for Promotion of Niigata University Research Projects.

Ancillary