Differential expression of parvalbumin in neonatal phencyclidine-treated rats and socially isolated rats

Authors

  • Sanne S. Kaalund,

    1. Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Copenhagen, Denmark
    2. Department of Neuroscience and Pharmacology, Protein laboratory, Copenhagen, Denmark
    Search for more papers by this author
    • These authors have contributed equally to the work of this article and therefore share the first authorship.
  • Jesper Riise,

    Corresponding author
    1. Synaptic Transmission 1, H. Lundbeck A/S, Valby, Denmark
    • Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Copenhagen, Denmark
    Search for more papers by this author
    • These authors have contributed equally to the work of this article and therefore share the first authorship.
  • Brian V. Broberg,

    1. Synaptic Transmission 1, H. Lundbeck A/S, Valby, Denmark
    2. Center for Neuropsychiatric Schizophrenia Research, Psychiatric Center Glostrup, Glostrup, Denmark
    Search for more papers by this author
  • Katrine Fabricius,

    1. Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Copenhagen, Denmark
    2. Synaptic Transmission 1, H. Lundbeck A/S, Valby, Denmark
    Search for more papers by this author
  • Anna S. Karlsen,

    1. Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Copenhagen, Denmark
    2. Department of Neuroscience and Pharmacology, Copenhagen, Denmark
    Search for more papers by this author
  • Thomas Secher,

    1. Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Copenhagen, Denmark
    Search for more papers by this author
  • Niels Plath,

    1. Synaptic Transmission 1, H. Lundbeck A/S, Valby, Denmark
    Search for more papers by this author
  • Bente Pakkenberg

    1. Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Copenhagen, Denmark
    Search for more papers by this author

Address correspondence and reprint requests to Jesper Riise, Laboratory of Origin, Research Laboratory for Stereology and Neuroscience, Bispebjerg University Hospital, Bispebjerg Bakke 23, DK-2400, Copenhagen NV, Denmark. E-mail: Jesperlriise@gmail.com

Abstract

Decreased parvalbumin expression is a hallmark of the pathophysiology of schizophrenia and has been associated with abnormal cognitive processing and decreased network specificity. It is not known whether this decrease is due to reduced expression of the parvalbumin protein or degeneration of parvalbumin-positive interneurons (PV+ interneurons). In this study, we examined PV+ expression in two rat models of cognitive dysfunction in schizophrenia: the environmental social isolation (SI) and pharmacological neonatal phencyclidine (neoPCP) models. Using a stereological method, the optical fractionator, we counted neurons, PV+ interneurons, and glial cells in the medial prefrontal cortex (mPFC) and hippocampus (HPC). In addition, we quantified the mRNA level of parvalbumin in the mPFC. There was a statistically significant reduction in the number of PV+ interneurons (= 0.021) and glial cells (= 0.024) in the mPFC of neonatal phencyclidine rats, but not in SI rats. We observed no alterations in the total number of neurons, hippocampal PV+ interneurons, parvalbumin mRNA expression or volume of the mPFC or HPC in the two models. Thus, as the total number of neurons remains unchanged following phencyclidine (PCP) treatment, we suggest that the decreased number of counted PV+ interneurons represents a reduced parvalbumin protein expression below immunohistochemical detection limit rather than a true cell loss. Furthermore, these results indicate that the effect of neonatal PCP treatment is not limited to neuronal populations.

Abbreviations used
CA

Cornu ammonis

CE

Coefficient of error

Cg1

Cingulate cortex 1

Cg2

Cingulate cortex 2

Cg3

Prelimbic cortex

CV

Coefficient of variance

DG

Dentate gyrus

GH

Group-housed

HPC

Hippocampus

HRP

Horseradish peroxidase

IHC

Immunohistochemical

IL

Infralimbic cortex

mPFC

Medial prefrontal cortex

neoPCP

Neonatal phencyclidine

NeuN

Neuronal nuclei

PCP

Phencyclidine

PND

Postnatal day

PV

Parvalbumin

ROI

Region of interest

SI

Social isolation

Vref

Reference volume

Schizophrenia is a chronic and disabling brain disorder characterized by heterogeneous positive and negative symptoms, as well as cognitive deficits that often manifest during adolescence. The inconsistencies in pathological findings (Pearlson 2000) and the increasing reports of genetic and environmental risk factors have led to the hypothesis that neurodevelopmental dysfunction and aberrant pruning result in abnormal brain connectivity in schizophrenia (McGrath et al. 2003). The affected neuronal circuits are believed to include prefrontal cortex, hippocampus and thalamus (Harrison and Weinberger 2005), structures that are interconnected in macrocircuits important for information processing.

The functionality of these macrocircuits is dependent on neurotransmitters including glutamate and GABA (Lisman et al. 2008). Post-mortem studies consistently show decreased expression of GABAergic biomarkers such as parvalbumin (PV) and glutamic acid decarboxylase 67 (GAD67) in schizophrenic brains (Curley et al. 2011; Beasley and Reynolds 1997; Hashimoto et al. 2003; Woo et al. 2004). Alterations in GABAergic signalling machinery are associated with cognitive impairments and may be a core pathophysiologic mechanism in schizophrenia (Rao et al. 2000; Constantinidis et al. 2002). The glutamatergic NMDAR antagonist phencyclidine (PCP) induces a psychosis-like state in healthy people and exacerbates symptoms in schizophrenia patients (Javitt and Zukin 1991; Olney and Farber 1995). In animal models, acute administration of PCP decreases PV expression (Cochran et al. 2003; Braun et al. 2007; Morrow et al. 2007), suggesting that NMDAR antagonists act most efficiently on NMDARs expressed by GABAergic interneurons (Olney and Farber 1995; Grunze et al. 1996). In addition, we and others have shown that neonatal PCP administration to rodents induces cognitive deficits relevant to schizophrenia in the adult animals (Broberg et al. 2008; Secher et al. 2009; Boctor and Ferguson 2010). The early postnatal period in rats corresponds to the second trimester of pregnancy in humans (Clancy et al. 2001), a period where especially viral infections increases the risk of developing schizophrenia later in life (Fatemi and Folsom 2009).

Another neurodevelopmental animal model of schizophrenia is social isolation of prepubescent rats, which predominantly induces psychotic-like symptoms associated with a subcortical hyperdopaminergic release in adult animals (Cilia et al. 2001; Fabricius et al. 2010). Furthermore, evidence of cognitive deficits and alterations of glutamatergic and GABAergic markers has also been found in this model (Fone and Porkess 2008). However, the mechanisms by which social isolation and neonatal NMDAR antagonism affects brain development and induces long-lasting cognitive deficits relevant to schizophrenia are unclear.

In this study, we examined putative long-term alterations of interneuronal PV expression in the two neurodevelopmental animal models with demonstrated cognitive deficits: neonatal PCP-treated rats (neoPCP) and socially isolated rats (SI).

Materials and methods

SI model

The SI paradigm has been described previously (Fabricius et al. 2010). Briefly, litters of Lister Hooded rats were cross-fostered on post-natal day (PND) 0 (Harlan, Venray, Netherlands) and arrived at an in-house facility on PND 8–9 with their foster mothers. Dams and pups were left undisturbed until PND 25, when the pups were weaned and randomly re-housed either singly (SI rats; n = 12) or in groups of 5 (group-housed controls (GH); n = 12). The SI/GH animal model was routinely used at H. Lundbeck A/S; thus, the surplus GH animals were submitted to other studies. Handling and noise were kept to a minimum during weekly cage cleaning. Rats were subjected to behavioural testing (see below) after 8 weeks of isolation, but were not killed until 15 weeks of isolation. A total of 12 SI male rats and 12 GH male rats were killed. Four GH rats were excluded from the study due to large shrinkage of the sections in the z-plan, bringing the final group sizes to 8 GH and 12 SI rats.

NeoPCP model

The neoPCP paradigm has been described in previous studies (Broberg et al. 2008). Briefly, timed-pregnant Lister Hooded rats (Charles River, Maidenhead, UK) arrived at an in-house facility on gestational day 14–15 and were housed individually. The day of parturition was noted as PND 0. Pups were cross-fostered and randomly assigned to lactating dams on PND 7. They were weaned on PND 25 and re-housed in groups of 2. Vehicle (0.9% isotonic saline; n = 24) or PCP (20 mg/kg PCP salt; n = 25) was administered subcutaneously to the pups on PND 7, 9 and 11. Rats were subjected to behavioural testing (see below) on week 8. A total of 24 vehicle-treated male controls and 25 neoPCP male rats were sacrificed. Twelve rats from each group were used for stereological quantification. Of these, 3 neoPCP rats were eliminated from the experiment due to the formation of ice-crystal artefacts in the tissue following storage in cryoprotection, bringing the final group sizes for stereological quantification to 12 controls and 9 neoPCP rats. Five controls and 5 neoPCP animals were used for in situ hybridization; 7 controls and 8 neoPCP rats were used for quantitative RT-PCR.

Behavioural testing

Locomotor activity was tested as described previously (Fabricius et al. 2010). Briefly, rats were observed in a novel test cage (Macrolon type III) for 60 min. The cage was equipped with 2 rows of 4 infrared light sources and photocells. Activity counts were recorded only when adjacent light beams in the lower row were interrupted, thereby preventing movement by a stationary rat from being counted as activity. Detection of locomotor and rearing activity was fully automated using custom-designed hardware and software (Ellegaard systems A/S, Faaborg, Denmark).

Tissue preparation (stereological quantification)

Rats were deeply anesthetized and perfused intracardially at 20 mL/min with 0.9% heparinized saline followed by 4% paraformaldehyde (w/v in 0.1 M phosphate-buffered saline). Brains were removed quickly and post-fixed for 48 h in 4% paraformaldehyde before being cryopreserved in 30% sucrose (w/v in 0.1 M phosphate-buffered saline) at 4°C until saturation. At the time of sectioning, whole brains were embedded in Tissuetek® (Sakura Finetek, Zoeterwoude, the Netherlands) and cut into 60-μm coronal sections (proceeding through the brain from anterior to posterior) using a Vibratome Ultrapro 5000 Cryostat (GMI Inc., Ramsay, Minnesota, USA). Sections were collected following stereological systematic random sampling principles (SURS) and stored individually in wells of cryoprotectant at −20°C.

Immunohistochemistry

Free-floating 60-μm sections were incubated with primary antibodies for 2 days followed by a 2-day incubation with the secondary antibody EnVision + horseradish peroxidase (HRP)-labelled polymer anti-mouse (Dako Nordic A/S, Glostrup, Denmark; working dilution 1 : 10). The following primary antibodies were used: mouse anti-neuronal nuclei (NeuN) monoclonal antibody (Milipore, Hellerup, Denmark; working dilution 1:18 000) and mouse anti-PV monoclonal antibody (Sigma-Aldrich, Broendby, Denmark; working dilution 1 : 10 000). Sections were mounted on Superfrost® Plus glass slides (Menzel, Braunschweig, Germany), dried and stained with cresyl violet (Sigma-Aldrich) before being dehydrated and coverslipped (Hounisen, Aarhus, Denmark).

Region of interest (ROI) delineation

The medial prefrontal cortex (mPFC) and the hippocampus (HPC) were delineated using the NewCast Software (Visiopharm, Hoersholm, Denmark) and based on the Paxinos and Watson rat brain atlas (Paxinos and Watsons 1986). The boundaries of the mPFC extended from the emergence of the forceps minor of the corpus callosum (bregma +3.7 mm) to the section before the decussation of the corpus callosum (bregma +1.2 mm) and included the anterior cingulate cortex, areas 1 and 2 (Cg1 and Cg2); the infralimbic cortex (IL); and the prelimbic cortex (Cg3) (Fig. 1a). The HPC was subdivided into the cornu ammonis (CA) and the dentate gyrus (DG), which were counted separately. The CA and the DG were delineated based on their morphology; both regions are characterized by high neuronal density compared to the surrounding tissue. The boundaries of the HPC were determined by the appearance of the CA and the DG anteriorly (bregma −1.80 mm) and the disappearance of the DG posteriorly (bregma −6.72 mm). The CA-subiculum boundary was selected at the level of the rhinal fissure.

Figure 1.

Delineation and sampling of the regions of interest (ROI) (a, b) and differentiation criteria for cell types (c, d, e) in immunohistochemically processed, frozen 60-μm sections. (a) The medial prefrontal cortex (mPFC) structure comprised the anterior cingulate cortex, area 1 (Cg1) and area 2 (Cg2); prelimbic cortex (Cg3) and infralimbic cortex (IL). (b) The hippocampal structure from the anterior to the posterior boundary. (c–d) NeuN specific staining, with arrows indicating neurons (N), astrocytes (A), oligodendrocytes (O) and endothelial cells (E). (e) Parvalbumin (PV)-specific staining where the arrow indicates a PV+ interneuron.

The stereological design

Total number of neurons, glial cells and PV+ interneurons, N, were estimated using the optical fractionator sampling design as described elsewhere (Fabricius et al. 2008) (the stereological sampling schemes for this study are summarized in Table S1). Estimates of the reference volume, Vref, for mPFC and HPC were obtained by employing Cavalieri's point-counting method combined with systematic random sampling (Gundersen and Jensen 1987; Gundersen et al. 1999). The precision of the estimates N and Vref is described by the coefficient of error, CE, which is the sampling error related to counting noise, systematic uniform random sampling and variances in section thickness (Gundersen et al. 1999; Jelsing et al. 2006;Riise and Pakkenberg 2011) (see Table S2 for an example of how the stereological estimates were calculated for this study). The counting procedure was performed using the NewCast software (Visiopharm, Hoersholm, Denmark) and a Nikon Eclipse 60i microscope (Nikon Nordic AB, Copenhagen, Denmark). The microscope was equipped with a Heidenhain electronic microcator measuring the z-axis and a ProScan II motorized stage system (Prior Scientific Instruments Ltd., Cambridge, UK) monitoring the x–y position. Slides were analysed using a 100X oil immersion objective with a high numerical aperture (NA = 1.4) and immersion medium (Olympus, Ballerup, Denmark). Digital live microscope images were visualized by a high-resolution Olympus DP72 camera (Olympus), with a final magnification of 2400X.

Counting criteria

Neurons and GABAergic PV+ interneurons were identified by immunohistochemical (IHC) staining using NeuN and PV antibodies (Fig. 1b). Glial cells and endothelial cells were distinguished based on cell morphology using a cresyl violet counter-staining of the Nissl substance in the cell nucleus. We did not attempt to differentiate glial cell types because there was significant variability in the morphological appearance of the various subtypes, making differentiation too uncertain.

Statistical analysis

Statistical analyses were performed using the SigmaPlot 11.0® (Alfasoft AB, Gothenburg, Sweden). Differences between animal groups were tested using an unpaired Student's t-test with a significance level of 0.05 (two tailed). If normality (Kolmogorov–Smirnov test) or equal variance (Levene-Median test) tests failed, non-parametric Mann–Whitney rank sum tests (two tailed) were performed. If nothing is indicated, the Student's t-test is used.

Radiochemical in situ hybridization

Rats were decapitated and the brains were removed quickly and frozen rapidly on dry ice prior to being stored at −80°C. Using a cryostat, 16-μm coronal sections were taken and thaw-mounted on Superfrost Plus® slides (Menzel, Braunschweig, Germany). For detection of PV mRNA in the mPFC, three sections per rat were hybridized as described in Rath et al. (2007) with a [35S]dATP-labelled 38 nucleotide antisense DNA probe corresponding to residues 115-153 of the rat PV mRNA (GenBank Accession no. NM_022499.2) (5′-ACCTTCTTCACATCATCCGCACTCTTTTTCTTCAGGCC-3′) or a sense control probe.

The hybridized sections were exposed to X-ray film for 25 days, and the film was developed. The autoradiographic images were digitized, and the optical density of the hybridization signals was quantified using ScionImage 1.42 software (Wayne Rasband, NIH, Bethesda, MD, USA). The optical densities were converted to dpm/mg tissue using simultaneously exposed 14C standards. The PV mRNA level in the mPFC region was quantified at the level of bregma +3.20 mm based on the rat brain atlas of Paxinos and Watson (1986).

Results

Social isolation increased locomotor activity in adulthood

To investigate the effect of neontal PCP administration and early social isolation on behaviour in adulthood, we assessed horizontal locomotor activity in a novel environment. We observed a significant increase in the activity level of SI compared with GH rats (< 0.001) (Fig. 2a), but no significant difference between neoPCP and control rats (= 0.76) (Fig. 2b).

Figure 2.

Rat locomotor activity as measured by photobeam breaks (activity counts) in 5-min time bins. The data are presented as mean activity counts (SEM) every 5 min over a period of 1 h. The bar graph shows the mean activity (SEM) during the first 60 min. In (a), group-housed (GH) rats (= 8) are indicated by open symbols and social isolation (SI) rats (= 12) are indicated by closed symbols. A significantly increased locomotor activity was observed in SI animals (< 0.001). In (b), neonatal phencyclidine (neoPCP) rats (= 12) are indicated by closed symbols and controls (= 12) are indicated by open symbols. No differences in locomotor activity were observed between control and neoPCP animals (= 0.76). *< 0.001.

Neonatal PCP reduced the number of PV interneurons and glial cells in the mPFC

Next, we counted the number of neurons, PV+ interneurons, and glial cells in the mPFC and HPC of the SI and neoPCP rat models (Fig. 3) using the fractionator method. PV+ interneurons were distinguished from other neuronal subtypes through IHC staining, whereas neurons and glial cells were distinguished from one another based on morphology. The stereological quantification of neurons and glial cells was limited to the mPFC. Interpretation of the following stereological data should be done with caution as CE in some cases contributed with more than 50% of the coefficient of variance (CV).

Figure 3.

Quantification of neurons, parvalbumin-positive (PV)+ interneurons, and glial cells in the medial prefrontal cortex (mPFC) and the hippocampus (HPC) (CA and DG), and PV mRNA expression. (a–e) Horizontal bars indicate group means. (a) The total number of neurons in the mPFC. No significant differences were observed in any of the two models (pGH-SI = 0.75, pControl-neoPCP = 0.59, two-tailed Mann–Whitney rank sum test). (b) The number of PV+ interneurons in the mPFC. There was a significant decrease in PV+ interneurons in neonatal phencyclidine (neoPCP)-treated animals (pControl-neoPCP = 0.021), but no difference in social isolation (SI) versus group-housed (GH) animals (pGH-SI = 0.94). (c) The number of glial cells. There was a significant decrease in glial cells in neoPCP-treated animals (pControl-neoPCP = 0.024) and a trend towards a decreased number of glial cells in SI animals (pGH-SI = 0.070, two-tailed Mann–Whitney rank sum test). (d) The number of PV+ interneurons in the CA (HPC) (pGH-SI = 0.38; pControl-neoPCP = 0.38). No significant differences were observed in any of the two models. (e) The number of PV+ interneurons in the DG (HPC). No significant difference was observed between SI and GH animals (pGH-SI = 0.88), but a trend towards an increased number was seen in neoPCP animals (pControl-neoPCP = 0.080). PV = Parvalbumin. *< 0.05 versus the appropriate control group. (f) Densitometric quantification of PV gene expression in the mPFC of neoPCP (= 5) and control (= 5) rats at PND 80. The data are presented as means (SEM) (= 0.76). On the right, an autoradiograph of the PV in situ hybridization is shown. Scale bar, 1 mm.

NeoPCP model

We observed a significant reduction in the number of PV+ interneurons bilaterally in the mPFC following neoPCP treatment (= 0.021) (Fig. 3b), but did not find statistically significant differences in the CA (p = 0.38), or the DG (= 0.080) of the HPC (Fig. 3d and e). However, a trend towards an increased number of PV+ interneurons was observed in the DG of the HPC. Furthermore, a significant reduction in the total number of glial cells in the mPFC was observed following neoPCP treatment (= 0.024) while the total number of neurons was unchanged (= 0.59 two-tailed Mann–Whitney rank sum test). Finally, we observed a trend towards a decrease in the mPFC volume in neoPCP rats (p = 0.076), but no differences in the volume of the HPC subregions DG (= 0.85) or CA (= 0.74). Total numbers and volumes are presented in Tables 1 and 2.

Table 1. Bilateral mean total number of PV+ interneurons, neurons and glial cells in the mPFC, HPC-CA and HPC-DG of the GH/SI and control/neoPCP model
 mPFC neuronsmPFC PVmPFC glial cells
MeanRangeCECV p MeanRangeCECV p MeanRangeCECV p
GH7.71 × 105[6.43; 11.3]0.050.200.754.95 × 104[2.45; 7.27]0.050.290.944.97 × 105[4.14; 5.76]0.060.100.070
SI7.47 × 105[5.58; 10.6]0.070.21 5.00 × 104[3.04; 6.64]0.070.21 4.28 × 105[ 4.21; 5.44]0.080.18 
Control8.81 × 105[7.35; 11.7]0.060.150.595.03 × 104[3.21; 6.59]0.070.200.0214.45 × 105[3.62; 4.65]0.080.150.024
neoPCP8.85 × 105[8.34; 10.3]0.060.07 4.10 × 104[2.91; 4.49]0.080.13 3.80 × 105[3.03; 4.63]0.080.13 
 HPC-CA PVHPC-DG PV
MeanRangeCECV p MeanRangeCECV p
  1. CE, Coefficient of error; CV, Coefficient of variance; PV, Parvalbumin; GH, Group housed; SI, Socially isolated; Control, Saline treated; neoPCP, Neonatal phencyclidine treated; mPFC, Medial prefrontal cortex; HPC, Hippocampus; CA, Cornu ammonis; DG, Dentate gyrus.

GH1.57 × 104[1.22;2.03]0.050.240.380.74 × 104[0.48;1.04)0.070.160.88
SI1.77 × 104[1.23;2.18]0.050.270.68 × 104[0.63;0.84]0.080.34
Control2.03 × 104[1.40;3.20]0.070.330.380.83 × 104[0.62;1.00]0.100.180.080
neoPCP1.75 × 104[0.89;2.58]0.070.321.14 × 104[0.66;1.52]0.090.39
Table 2. Bilateral mean volume of mPFC, HPC-CA and HPC-DG (mm3) in the GH/SI and control/neoPCP model
 mPFCHPC-CAHPC-DG
MeanRangeCECV p MeanRangeCECV p MeanRangeCECV p
  1. CE, Coefficient of error; CV, Coefficient of variance; PV, Parvalbumin; GH, Group housed; SI, Socially isolated; Control, Saline treated; neoPCP, Neonatal phencyclidine treated; mPFC, Medial prefrontal cortex; HPC, Hippocampus; CA, Cornu ammonis; DG, Dentate gyrus.

GH8.22[6.93; 9.55]0.010.110.0767.44[6.33; 0.95]0.020.160.324.42[3.70; 4.74]0.020.100.69
SI7.42[5.61; 9.05]0.020.136.69[6.00; 7.59]0.010.104.06[3.49; 5.15]0.020.18
Control8.05[6.59; 10.5]0.010.120.0766.78[5.77; 8.08]0.020.120.743.90[3.40; 4.47]0.030.080.85
neoPCP7.39[6.82; 8.02]0.020.056.93[6.06; 8.41]0.020.143.94[3.52; 4.76]0.030.12

In summary, neonatal NMDAR blockade led to an 18% reduction in the number of IHC-detectable PV+ interneurons in the mPFC, but not in the HPC of adult rats. However, a trend towards an increased number of PV+ interneurons was observed in the DG of HPC following neoPCP treatment. The total number of neurons in the mPFC was unaltered. NMDAR antagonism also led to a 15% reduction in the number of glial cells. No differences between control and neoPCP rats were observed in the regional volumes though a trend towards a reduced mPFC volume was present in neoPCP animals.

SI model

We found no differences in the number of PV+ interneurons (= 0.94) or total number of neurons (= 0.75) in the mPFC between SI and GH rats (Fig. 3a and b). On the basis of the observation of unchanged PV+ interneurons in the mPFC, we limited the stereological quantification of PV+ interneurons to 4 animals in each group for the HPC. As hypothesized, we found no significant differences in PV+ interneurons in the CA (= 0.38) or DG (= 0.88) of the HPC (Fig. 3d and e). We observed a trend towards a decreased total number of glial cells (= 0.070, two-tailed Mann–Whitney rank sum test) (Fig. 3c) and a reduced volume (p = 0.076) of the mPFC following social isolation, but no significant differences in the volume of the DG (= 0.69) or the CA (= 0.32) of the HPC. Total numbers and volumes are presented in Tables 1 and 2.

In summary, we found that social isolation did not cause any significant alteration in the total number of neurons, PV+ interneurons or glial cells, nor did it lead to a significant volume reduction in the mPFC or the HPC (CA or DG). However, there was a trend towards a reduced mPFC volume and a reduction of glial cells in SI rats compared with controls.

Neuronal density

When correcting the total number of neurons in the mPFC for volumetric differences, we observed no significant differences in neuronal density between control and neoPCP rats (= 0.064). However, a trend towards an increased neuronal density was observed in the mPFC of the neoPCP animals. Furthermore, we also observed a strong trend towards a decreased glial density in the mPFC following neoPCP treatment (= 0.053). We failed to see a reduction in PV density of neoPCP animals in the mPFC (= 0.14), as would have been expected from the cell loss; however, this could be masked by the trend level decrease in the mPFC volume. Finally, we did not find any significant changes in PV density of the CA (p = 0.19) or the DG (= 0.085) of the HPC in the neoPCP model. Nevertheless, a trend towards an increased PV density was observed in the DG of the HPC in neoPCP animals.

We did not observe any significant changes in neuronal density (= 0.49), PV density (= 0.39) or glial density (= 0.34, two-tailed Mann–Whitney rank sum test) in the mPFC of the SI model nor did we find any significant changes in PV density of the CA (= 1.0) or the DG (= 0.49) of the HPC between GH and SI rats. The density values are presented in Table 3.

Table 3. Bilateral mean density (cells/mm3) of neurons, PV+ interneurons and glial cells in mPFC, HPC-CA and HPC-DG in the GH/SI and control/neoPCP animals
 mPFC NeuronsmPFC PVmPFC glial cellsHPC- CA PVHPC-DG PV
MeanRange p MeanRange p MeanRange p MeanRange p MeanRange p
  1. PV, parvalbumin; GH, group housed; SI, socially isolated; Control, saline treated; neoPCP, neonatal phencyclidine treated; mPFC, medial prefrontal cortex; HPC, hippocampus; CA, cornu ammonis; DG, dentate gyrus.

GH9.68 × 104[8.71; 11.7]0.496.00 × 103[2.85; 7.89]0.396.16 × 104[5.54; 7.91]0.340.24 × 104[0.18; 0.29]1.00.16 × 104[0.13; 0.19]0.49
SI10.1 × 104[7.74; 12.2]6.61 × 103[5.32; 7.68]5.86 × 104[4.86; 8.29]0.24 × 104[0.19; 0.29]0.18 × 104[0.13; 0.24]
Control11.0 × 104[9.45; 12.1]0.0646.24 × 103[4.80; 8.88]0.145.66 × 104[4.53; 6.61]0.0530.30 × 104[0.21; 0.44]0.190.21 × 104[0.14; 0.27]0.085
neoPCP11.7 × 104[10.6; 12.8] 5.57 × 103[3.87; 7.22]5.04 × 104[4.23; 5.91]0.25 × 104[0.15; 0.31]0.29 × 104[0.15; 0.42]

Neonatal PCP treatment did not alter mRNA expression in adulthood

To determine whether the reduced number of PV+ interneurons in the mPFC was accompanied by reduced transcription of the PV gene, we performed radioactive in situ hybridization and quantitative RT-PCR. The specific hybridization pattern of the PV antisense probe is consistent with previously published studies (Romon et al. 2011; Seto-Ohshima et al. 1989) and is in agreement with the distribution of PV immunoreactivity (Hof et al. 1999). As measured by densitometry, there was no long-term effect of the neoPCP treatment on PV mRNA expression in the mPFC (= 0.76, ncontrol = 5, nneoPCP = 5) (Fig. 3f). This finding was supported by quantitative RT-PCR (= 0.82, ncontrol = 7, nneoPCP = 8, data not shown).

Discussion

Cognitive deficits are central features of schizophrenia. Coordinated synchronization among brain regions is important for cognitive processing (Uhlhaas and Singer 2006). Several studies have suggested that gamma frequency oscillations contribute to the control of cognitive performance by entraining a strong and synchronous inhibition of networks of pyramidal neurons (Sohal et al. 2009). Thus, deficient GABA neurotransmission could lead to cognitive impairment.

Recent optogenetic studies have established that fast-spiking GABAergic PV+ interneurons drive cortical gamma oscillations in mice (Sohal et al. 2009). Our understanding of the importance of the PV+ interneurons is increasing; however, many studies have relied on qualitative observations as well as density measurements of cells, leading to inconsistent findings. This study is the first to examine the number of PV+ interneurons in the mPFC and the HPC of adult rats based on stereological principles. Our findings concern the quantitative, long-term effect of SI and neonatal PCP treatment on the total number of neurons, PV interneurons, and glial cells in the mPFC and the HPC, and on the regional volume of these structures.

NeoPCP model

We used the optical fractionator method to estimate the total number of IHC-stained, PV+ interneurons. We found that the PV+ interneuron population in the mPFC comprised 4.10 × 104 cells. Moreover, we observed a persistent 18% loss of IHC-stained PV+ interneurons in the adult mPFC. In contrast, we found a weak trend towards an increase in total number and density of PV+ interneurons in the DG subregion of the HPC, while no changes in the combined CA1-4 regions were observed. Regional differences in developmental progression and unknown tolerance mechanisms may explain the differential susceptibility to early neoPCP treatment of the mPFC and the HPC (Swann et al. 1989; Cochran et al. 2003).

The PV protein is expressed specifically in fast-spiking interneurons in the mPFC and the HPC. Reductions in PV are generally believed to have an impact on neuronal functionality, due to altered Ca2+ buffering capacities (Sohal et al. 2009). However, it is unclear whether the observed loss of IHC-stained PV+ interneurons reflects decreased expression of the PV protein or a loss of cells. This issue was specifically addressed in a study by Powell et al. 2012; where the expression of Pv-promotor driven green fluorescent protein was compared with the PV protein level in mice. They reported no PV-Green fluorescence protein specific cell loss, but described a decreased expression of the PV protein. These findings suggest that the development, but not survival, of PV+ interneurons is hampered by neonatal NMDAR blockade. Consistent with this, we observed a decrease in the number of IHC-stained PV+ interneurons but no change in PV expression at the mRNA level in the mPFC. Thus, our results indicate that the PV deficit following NMDAR blockade originates at the post-transcriptional level. PCP does not seem to cause an overall induction of neurodegeneration, as the total number of neurons was unchanged. However, it should be stressed that we did observe a trend towards an increased neuronal density in the neoPCP-treated animals, though this most likely relates to the trend level decrease in mPFC volume. Thus, our results suggest that GABAergic interneurons are more susceptible to pathological changes due to NMDAR antagonism than other neuronal cell types (Olney and Farber 1995; Grunze et al. 1996).

We were unable to establish a significant difference in the PV density of the mPFC, which may be related to the trend level decrease in the mPFC volume.

We also found a significant 15% decrease in the number of glial cells and a strong trend towards decreased glial density in the mPFC of neoPCP rats. Because we were unable to distinguish between glial subtypes based on morphology, our data are silent as to whether there was a homogenous glial cell loss or if the loss was restricted to certain subtypes. Previous studies indicate that NMDARs are expressed by astrocytes (Krebs et al. 2003) and oligodendrocytes (Karadottir et al. 2005), but the effect of NMDAR blockade on glial cells is inconclusive (Hajszan et al. 2006; Kondziella et al. 2006; Lindahl et al. 2008). Across studies, NMDAR antagonism has generally led to alterations of glial cells, consistent with our findings.

SI model

We found no changes in the number or density of PV+ interneurons, neurons or glial cells in the mPFC or HPC following social isolation. However, it should be noted that a trend towards a reduction in the total number of glial cells and a volume decrease in the mPFC was observed in SI animals. Our findings are consistent with those of an earlier stereological study of neuron number and density (Day-Wilson et al. 2006), confirming the robust nature of stereological estimations. In contrast, non-stereological studies have reported down-regulation in SI animals of PV+ interneurons in the DG of the HPC, the CA 2/3 of the HPC and the PFC (Harte et al. 2007); reduction of chandelier cartridges in the prelimbic cortex (Bloomfield et al. 2008); and decreased synaptic content in the DG of the HPC (Varty et al. 1999). Our findings do not confirm these reported alterations in the number of PV interneurons.

Behavioural observations

In this and a previous study, we observed hyperactivity in the SI model upon exposure to a novel environment (Fabricius et al. 2010). Previously, this hyperactivity has been associated with a hypodopaminergic mPFC and increases in subcortical dopamine transmission, thus mimicking core features of schizophrenia (Robbins et al. 1996; Heidbreder et al. 2000). Increased locomotion has also been observed in rats subjected to neonatal NMDA antagonism when challenged by PCP exposure in adulthood but not at baseline (Abekawa et al. 2007; Fabricius et al. 2010). However, the increased locomotion is more likely driven by a reduction in PV interneuron-related GABA transmission rather than dopaminergic deficits in this animal model. This assumption is supported by the reports of excessive glutamate efflux (Amitai et al. 2012), decreased markers of GABA function in PFC following PCP treatment (Adams and Moghaddam 1998) and increased firing rate of pyramidal neurons in the PFC following NMDA antagonism (Jackson et al. 2004).

Overall, our behavioural and stereological observations of the two models suggest that they model different aspects of schizophrenia pathology and symptomatology. The strong link between perturbation of the glutamatergic and GABAergic transmission in the PFC and cognitive function (Lewis and Moghaddam 2006) favours the use of the neoPCP model in future studies aiming to unravel the role of PV+ interneurons in cortical dysfunction relevant to cognition and schizophrenia. However, the SI model may reflect a different construct compared with the neoPCP model when it comes to modelling positive symptoms as SI induces hyperactivity and affects the dopaminergic system (Fabricius et al. 2010). Finally, our study is the first to quantify the total number of PV+ interneurons in the rat brain. We found that there are, on average, ~50 000 PV+ interneurons in the mPFC, ~20 000 PV+ interneurons in the CA1-4, and ~8300 PV+ interneurons in the DG of the HPC. The estimate of PV+ interneuron number in the mPFC and the HPC may be of particular relevance for modelling of cortical circuits where the relative proportions of different neuronal populations and network sizes are crucial for implementing electrophysiological data into the mathematical network models of cognition (Berg and Hounsgaard 2009). Understanding how cortical circuits function in the normal brain will aid to our understanding of the deficits underlying cortical dysfunction in schizophrenia and the development of treatment strategies hereof. Subsequent studies should be done to evaluate the functional outcome of reduced PV expression in neoPCP and identify pharmacology that could compensate for the PV deficit.

Acknowledgements

The authors declare no conflict of interest. We thank Helen Nielsen and Susanne Sørensen for their technical assistance. This project was approved by the Danish Regional Scientific Ethical Committee (protocol numbers: SI animals – 350-C16; neoPCP animals – 366-C23). Financial support was granted by the Copenhagen Graduate School of Health Science, University of Copenhagen and Aase and Ejnar Danielsens Foundation.

Authorship credit

SS Kaalund and J Riise have contributed equally to the work of this article; they have participated to the conception and design, acquisition and analysis of the stereological results, interpretation of data and drafting the article. B Broberg, K Fabricius and T Secher has performed the handling and testing of animals. A Karlsen has performed and analysed the mRNA quantification data. N Plath and B Pakkenberg have contributed to the conception and design of the study as well as critically revising the article. All authors have read and approved the submitted version of the article.

Ancillary