Prenatal immune activation induces age-related alterations in rat offspring: Effects upon NMDA receptors and behaviors

Prenatal exposure to polyriboinosinic-polyribocytidylic acid (poly I:C) results in psychotic-like behavior in mature rat offspring as well as enduring modifications of glutamatergic excitatory synaptic transmission. However, little is known about the dynamic behavioral and glutamate N-methyl-D-aspartate (NMDA) receptor changes in rat offspring following poly I:C treatment of pregnant dams. In this study, poly I:C was administered to rats intravenously at a dose of 10 mg/kg on gestational day 9 in order to assess changes in behavior and NMDA receptors in offspring over time. Results demonstrate progressive worsening behaviors in adolescents and adults that manifest as increased anxiety, cognitive impairment, and pre-pulse inhibition deficits. Age-related alteration of NMDA receptors in the prefrontal cortex and hippocampus, either total number or distribution, were observed from weaning to adulthood. These results suggest that abnormalities of NMDA receptors occur prior to obvious schizophrenia-like behavioral manifestations. Hence, NMDA receptors may be potential therapeutic targets to prevent disease development during asymptomatic periods of schizophrenia, and may serve as targets for preventive and/or therapeutic strategies for schizophrenia. Further, PSD95, a scaffolding protein that is a component of the NMDA receptor signaling complex, is increased in the hippocampus of adult offspring, when serious behavioral abnormalities emerge. This result suggests that PSD95 may be involved in behavioral abnormalities of schizophrenia.


Introduction
Ethological studies have demonstrated maternal immune activation (MIA) to induce inflammation and that viral or bacterial insult are risk factors for multiple neurodevelopmental disorders such as autism and schizophrenia [1][2][3][4]. One common approach to induce MIA in rats is to administer double-stranded RNA analogue polyriboinosinic-polyribocytidylic acid (poly I:C) to pregnant animals. Poly I:C induces cytokines associated with the immune response in maternal and fetal compartments, including the fetal brain [5,6], and is similar to that induced by viral infection [7]. Numerous studies have shown that the offspring of pregnant dams treated with poly I:C show a battery of schizophrenia-like behaviors and neurodevelopmental abnormalities [8][9][10]. Therefore, the poly I:C MIA model has become a powerful tool for the investigation of the progressive nature of schizophrenia and for the development of therapeutic and preventive strategies to halt disease progression [11].
Research has shown that differing time periods of MIA induction result in differing emergent behavioral and/or cognitive pathological symptom clusters during offspring life [12]. It was once thought that the mid -late prenatal period was the only risk window for schizophrenia development, which resulted in the emergence of perseverative abnormal behaviors [13]. However, several studies have demonstrated prenatal immunological stimulation during early gestation to induce psychiatric disease [14,15]. Animal research has shown that prenatal immunological stimulation during early gestation, especially at gestational day (GD) 9, specifically induces abnormal behaviors, such as, impaired working memory [16] and pre-pulse inhibition (PPI) deficits [17]. However, there is limited behavioral abnormalities data, regard to dynamic change over time, especially in early MIA induced by poly I:C on GD9.
N-Methyl-D-aspartate (NMDA) receptors are glutamate-gated, calcium-permeable ion channels that are of particular interest, in that they participate in synaptic transmission and are implicated in various processes, such as learning [18], memory [19], and long-term neuronal potentiation [20]. Dysfunction of NMDA receptors is thought to play a role in the pathophysiology of neurodevelopmental diseases like schizophrenia [21][22][23]. A genome-wide association study (GWAS) suggested that NMDA receptors may play a leading role in the pathology of schizophrenia, as they are highly targeted by multiple types of genetic risk factors [24]. Prenatal exposure to viral infection or MIA has been repeatedly shown to cause NMDA receptor abnormalities [25,26]. However, details about the dynamic changes in NMDA receptor expression early in the prenatal MIA rat model have not been well described and needs further exploration.
The aim of this study was to assess dynamic changes in behavior and NMDA receptors over time in early prenatal MIA offspring. The purpose was to evaluate the molecular influence of environmental factors on the central nervous system, providing clues for future development of prevention and/or therapeutic strategies for schizophrenia.

Animals
Thirty female and male Sprague-Dawley rats were obtained from Beijing Vital Rival (Beijing, China). Eight-week-old animals were derived from multiple litters. Littermates of the same sex were caged together, with 3-4 per cage. The animals were mated at 3 months of age, and the presence of vaginal plugs was used to confirm copulation. The procedures for breeding and for the verification of pregnancy have been described by Meyer [27]. The rats were housed individually in ventilated plastic cages in a temperature-and humidity-controlled (22 ± 2°C, 50 ± 10%) holding facility with a constant day -night cycle (lights: 08:00-20:00). All animals had ad libitum access to food and tap water. The Animal Care and Use Committee of the Henan Key Laboratory of Biological Psychiatry (Xinxiang, China) approved the use of the rats and the experimental protocols in this study.

Maternal immune activation during pregnancy
For maternal immunological manipulation during pregnancy, female rats were subjected to a timed mating procedure as described above. In this study, we defined the first day following copulation as gestation day (GD) 1. On GD 9, the dams were randomly divided into two groups (N = 14 for poly I:C and N = 15 for saline litters), which were given a single tail vein injection of poly I:C in saline at 10.0 mg/ kg, or saline equivalent [12,28]. All solutions were freshly prepared on the day of administration. Three randomly selected pregnant rats were humanely sacrificed approximately 3 h after the administration of poly I:C or saline. Five ml of blood were collected by cardiac puncture using sterile syringes. The blood was then centrifuged at 3000 × g for 15 min and maternal plasma stored at −80°C before measurement of interleukin (IL)-1β, IL-6, and tumor necrosis factor α (TNF-α). Subsequent procedures were all performed with highly sensitive enzyme-linked immunosorbent assays (ELISA) from company R&D Systems. Other animals were immediately returned to their home cages after injection.

Allocation and testing of offspring
Each pregnant rat provided on average 10 ± 2 offspring. On postnatal day (PND) 21, both groups of pups were weaned, and then group housed by sex and prenatal group (3-4 per cage using 1-2 rats per litter in experiments). Males (n = 12 control, n = 12 poly I:C) were cognitively evaluated as adolescents (PND35-45) and adults (PND56-65) as described below. Rats were excluded that did not complete all the tests. After completion of behavior testing at each developmental stage, animals were anesthetized using sodium thiopental (50 mg/kg, i. p.). The hippocampus (HIC) and prefrontal cortex (PFC) from each rat was assessed for gene and protein expression (n = 8 control, n = 8 poly I:C). Other animals were perfused with 4% paraformaldehyde and brains removed for immunofluorescent analysis (n = 4 control, n = 4 poly I:C). With the exception of behavioral testing, all procedures were also completed at PND21. The adolescent and adult stages were defined based on the gradual attainment of sexual maturity and age-specific behavioral discontinuities from younger to older animals [29]. (These two developmental stages roughly correspond to 11-14 years and to 20 years and older for humans.) Each brain was frozen immediately on dry ice before being transferred to storage at −80°C until required for analysis.

Behavioral testing
Each group of 12 males was randomly selected at PND40 (adolescent stage) and at PND60 (young adult stage) for standard behavioral evaluation of schizophrenia-like behavior. The Open Field Test was completed on the first day. The Elevated Plus Maze (EPM) was completed on day 2. These tests were used to investigate the level of anxiety and locomotor activity in experimental animals. The Y-maze task was completed on day 3 and the step-through passive avoidance test completed on day 4 and day 5. A standard behavioral protocol, the Barnes Maze Test, was completed between the 6th and 9th days to evaluate cognition, including working and spatial recognition memory. On the last day, a pre-pulse Inhibition (PPI) test was completed, in order to measure sensorimotor gating. All behavioral tests were performed between 08:00 and 18:00 h. Each rat was assessed for each behavioral measure. After each trial, the apparatus was cleaned with 75% alcohol.

Open field test
Locomotor activity was measured in a 100 cm × 100 cm open field arena (100 cm × 100 cm × 50 cm, length × width × height, respectively). Male rats underwent a 3-min adaptation and then a 5-min exploration period within the testing zone and were video-recorded using Spain Pan Lab Smart 3.0. The apparatus was cleaned between trials with a 75% alcohol solution. All experiments maintained the same experimental conditions.

Elevated plus maze test
The elevated plus maze test is an ethologically relevant assessment of anxiety levels in rodents. The EPM apparatus consisted of two open arms (10 × 60 cm 2 ) and two closed arms (10 × 60 × 40 cm 3 ), connected by a central platform (10 × 10 cm 2 ) and elevated 40 cm above the floor. Subjects were individually placed in the center of the maze facing a closed arm and allowed to roam freely for 5 min before return to their home cage with dim lighting. An anxiety index was calculated with the formula: The maze was cleaned with a 75% ethanol solution to remove odor trails.

Spatial novelty preference in the Y-maze
The Y-maze task is a simple test for measuring spatial recognition memory by spatial novelty preference. The apparatus consists of three identical black iron plate arms radiating from a central triangle and spaced 120°from each other, and identified as the start arm, familiar arm (with a red label), and novel arm (with a yellow label). The task consists of a sample phase and a choice phase. During the first phase, rats were; introduced at the start arm (facing the central triangle), allowed to explore the start arm and familiar arm for 10 min with the novel arm blocked by a door, and then returned to their home cage. The door was removed after 2 h so that all three arms were accessible and rats allowed to freely explore for 5 min. Entries to and time spent in each of the three arms were video-recorded and analyzed using Spain Pan lab Smart 3.0. Spatial recognition memory was analyzed by comparing the animal's exploration of all three arms.

2.4.4.
Step-through passive avoidance test Given that rats naturally prefer dark environments, a shuttle-box apparatus was used to measure inhibitory avoidance memory. The apparatus consisted of a bright chamber (25 × 25 × 20 cm) and dark chamber (25 × 25 × 20 cm) with grid door. The chambers were separated by a guillotine door. Electrical shocks (2 s and 0.5 mA intensity) were transferred by a standard stimulator to the grid floor in the dark chamber. The first day, each animal was initially placed in the bright compartment and the door between the two compartments was opened after 30 s. The initial latency for a rat to enter the dark compartment was measured (T1). Immediately after the movement of the animal to the dark compartment, the door was closed and an electric foot-shock (2 s, and 0.5 mA intensity) was provided through stainless-steel rods. If the latency of the animal was >300 s in the first acquisition trial, that animal was excluded from the study. On the second day the same paradigm was repeated, but without foot-shock. The step-through latency (T1) for entrance to the dark chamber was recorded on the first day and the second day (T2). Time latency was calculated using the formula (T2-T1), which represents the learning and memory ability of the rats.

Barnes Maze Test
We examined spatial learning and memory by use of the Barnes maze, which is elevated 140 cm above the floor and consists of 20 holes located evenly on the surface periphery, each of which is 5 cm in diameter. The target box was one of the holes that connected to a dark chamber, allowing the animal to escape from bright light exposure. The day before the formal experiment, animals were adapted in the target box for 4 min. On the first day each animal was placed in the center black cube of the maze for 5 s and permitted to explore the maze in order to find the target box, when the cube was removed. Once the animal entered the escape box, the animal was left there for 30 s, if it failed to find the target box within 3 min it was taken to the target box and allowed to remain in the target box for 1 min. Each animal underwent two trials during the day with an interval of 4 h between trials. The time of latency to reach the target box and the number of errors committed by each animal within 180 s were recorded. An error was defined as a head poke or exploration of any hole other than the hole above the target box, including perseverative investigations of the same hole. The test was performed on 4 days. Between tests, 75% alcohol was used to clean the Barnes maze to avoid olfactory cues. The whole process was monitored by a digital camera and a computer system.

Pre-pulse inhibition (PPI) test
Sensorimotor gating was assessed using the paradigm of PPI as an acoustic startle reflex. PPI was conducted in four sound-attenuated chambers. All test sessions were performed in a single-chamber startle apparatus (QMC-I, Kunming Institute of Zoology, Chinese Academy of Sciences, China). The white noise was set to 70 decibels (dB) for 10 pretests, adaptation was for 5 min and the formal experiment begun. Rats received five startle trials (120-dB bursts of white noise lasting 20 ms). The experiment consisted of 40 rounds of stimulation that were randomly distributed. First, a delay of˜50 ms, followed by 3 types of pre-pulse inhibition trials (75, 80, or 85 dB) for 20 ms (10 times each), followed by a 100 ms delay, and then a 40 ms stimulation of the startle reflex with white noise at 120 dB. The session closed with another five startle stimuli. The average interval was 15 s for all trials. The results, designated PP75, PP80, and PP85 were calculated automatically by system software. The percentage PPI induced by each pre-pulse intensity was calculated as [1 − (startle amplitude on pre-pulse trial/ startle amplitude on pulse alone)] × 100%.
PCRs were performed in a 20 μl volume with SYBR Green Master Mix (promega) and 1 μm pre-primer and 1 μM post-primer according to the recommendations of the manufacturer. PCR conditions were: 95°C for 30 s; for 40 cycles; 95 0 C for 5 s, 60 0 C for 34 s, and 72 0 C for 1 min. The relative mRNA expression levels of NR1, NR2A, NR2B and PSD95 were normalized to that of GAPDH and calculated using the ΔΔCt method [29].

Protein extraction and western blot analysis
Western blot analysis was carried out using the following primary antibodies raised against target proteins: NR1 (rabbit monoclonal, ab109182, 1:2000 dilution) NR2A (rabbit monoclonal, ab124913, 1: 1000 dilution), NR2B (rabbit polyclone, ab65783, 1: 500 dilution), and PSD95 (mouse monoclonal, ab13552, 1:1000 dilution). The following secondary horseradish peroxidase (HRP)-conjugated antibodies were used at 1: 5000 dilution: goat anti-rabbit HRP (12-348) (Millipore, Watford, UK); and goat anti-mouse (sc-2005) (Santa Cruz, Insight Biotechnology, Wembley, UK). Western blot analysis was used to investigate NMDA receptors and PSD95 protein levels in the total homogenate of target brain area tissues. Brain sample homogenates were prepared in radioimmunoprecipitation assay (RIPA) buffer and centrifuged at 12,000 × g for 15 min at 4 0 C. Supernatants were collected and protein concentration determined by the Bio-Rad Coomassie Blue protein assay (Bio-Rad, Hemel Hempstead, UK). Equal amounts of protein (40 μg) were assessed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) at 80 V for 100 min to separate proteins according to their molecular weight. The separated proteins were then blotted onto polyvinyl alcohol difluoride membranes (Life Technologies, Paisley, UK) at 300 mA for 90 min. Blots were blocked for 1 h in 5% non-fat milk, before overnight incubation at 4°C with the appropriate primary antibody (diluted in 5% milk-trisbuffered saline with Tween 20 (TBST). Membranes were then washed three times for 10 min with TBST and incubated with the appropriate horseradish peroxidase (HRP) conjugated secondary antibody (prepared in 5% milk-TBST) for 1 h at room temperature. Following secondary antibody incubation, blots were washed three times for 10 min with TBST before being visualized using an Enhanced Chemiluminescence Plus detection kit (GE RPN2232).

Immunofluorescent assessments
Brain tissues embedded in (OCT) were cut into 10 μm sections. After the antigens were treated with microwaves, slices were then washed twice with 0.1 M phosphate buffered saline (PBS), and incubated at room temperature for 30 min in 10% bovine serum albumin (BSA). The sections were incubated with rabbit monoclonal anti-NR1 (ab109182 1:100 dilution) overnight at 4 0 C. The next day, after washing the sections with PBS three times, secondary antibody conjugated to the fluorescent marker CY3 (1: 100; Bost Biotech, Wuhan, China) was added to each section and incubated at room temperature for 2 h, then, washed three times with PBS. Finally, slices were mounted with an antifluorescence quenching agent. Within all fluorescence images, the HIC and PFC were selected. Stained sections were imaged using a panoramic 250 slide scanner and data analysis software. Areas of immunofluorescence intensity were quantified with a computer-assisted image analysis program (ImageJ). The area of highlighted immunoreactivity was used to calculate the average fluorescence intensity of the selected area. Scale bar is shown as 100 μm.

Data analysis and statistics
The data are presented as means ± standard error of the mean (SEM) and were analyzed with SPSS 20.0 (SPSS Inc., Chicago, IL, USA). All gene expression (RT-PCR), cytokine levels, and protein expression (Western blot) data were analyzed using independent Student's t tests (two-tailed). Two-way analysis of variance (ANOVA) (treatment × age) was used for behavioral data analysis (Open Field Test, Elevated Plus Maze Test, Y-maze) and NR1 immunofluorescence analysis. Two-way ANOVA (treatment × trials) with repeated measures (RM) was used for behavioral data analysis (Barnes Maze Test and Pre-pulse Inhibition (PPI) Test), followed by Bonferroni correction for chosen group comparisons. Correlative analyses among behavioral measures and NMDA receptor protein levels were performed using first-order partial correlations. The probability level less than 0.05 was accepted as significant.

Cytokine levels of immune-challenged mothers
Poly I:C was administered at a dose of 10 mg/kg on gestational day 9 to female rats and the systemic inflammatory response was detected 3 h after poly I:C injection. The levels of IL-6, IL-1β, and TNF-α in the peripheral blood of pregnant rats are summarized in Fig. 1 (all p < 0.05). Increased cytokine levels were observed in immune-activated pregnant rat, confirming the efficacy of poly I:C.

Prenatal immune activation induces maturation-dependent alterations in offspring behavioral phenotypes
The functional consequences of immunostimulant treatment were evaluated in MIA offspring using a battery of behavioral tests at PND40 and PND60, which represent adolescence and adulthood, respectively. Two-way ANOVA and Two-way ANOVA with repeated measures were used for data analysis, followed by Bonferroni correction for chosen group comparisons Fig. 2.

Open field test
When MIA offspring was exposed to an open field arena for 5 min of exploration, two-way ANOVA showed that factor age effect (F (1,44) = 0.1603, p > 0.05), the main effect of factor age was insignificant; Factor treatment effect (F (1,44) = 2.089, p > 0.05), the main effect of factor treatment was insignificant; Factor age × factor treatment (F (1,44) = 55.58, p < 0.0001), the interaction effect between two factors was significant. Post-hoc Bonferroni showed that comparing to the control group, the total motor distance of MIA offspring was significantly increased in PND40 (p < 0.001) and significantly decreased in PND60 (p < 0.01) (Fig. 2.1 A). Two-way ANOVA showed that factor age effect (F (1,44) = 0.1290, p > 0.05), the main effect of factor age was insignificant; Factor treatment effect (F (1,44) = 2.059, p > 0.05), the main effect of factor treatment was insignificant; Factor age × factor treatment (F (1,44) = 10.06, p > 0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that comparing to the control group, there was no difference in central area at PND40 (p > 0.05), but an increased trend was observed, and distance moved in the central area was decreased at PND60 (p < 0.01) (Fig. 2.1  B). When MIA offspring was exposed to an open field arena, there was no difference in the total motor distance for 30 min of exploration between MIA offspring and control at PND60 (F (1,44) = 4.268, p>0.05) (Fig. Supplement 2A).

Elevated plus maze test
In the elevated plus maze test, two-way ANOVA showed that factor age effect (F (1,44) = 0.6101, p>0.05), the main effect of factor age was insignificant; Factor treatment effect (F (1,44) = 0.2090, p>0.05), the main effect of factor treatment was insignificant; Factor age × factor treatment (F (1,44) = 19.53, p < 0.0001), the interaction effect between two factors was significant. Post-hoc Bonferroni showed that comparing to the control at PND60, prenatal poly I:C treatment decreased entries into open arms(p < 0.01) and increased at PND40 (p < 0.05) (Fig. 2.1  C). Two-way ANOVA showed that factor age effect (F (1,44) = 1.643, p>0.05), the main effect of factor age was insignificant; Factor treatment effect (F (1,44) = 0.0008297, p>0.05), the main effect of factor treatment was insignificant; Factor age × factor treatment (F (1,44) = 0.06587, p>0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that comparing to the control, there was no difference was found on the time spent in open arms at PND60 and PND40 (p > 0.05) (Fig. 2.1 D).
These results suggest that poly I:C treated offspring display increased locomotor activity and lower innate anxiety in adolescence, whereas in adulthood anxiety-related behavior moderately increased. In order to estimate memory impairment, the core symptom of schizophrenia, the Y-maze test and the Barnes maze test, were carried out.

Spatial novelty preference in the Y-maze
During a 5 min period in the Y-maze test, two-way ANOVA showed that factor age effect (F (1,44) = 0.8341, p>0.05), the main effect of factor age was insignificant; Factor treatment effect (F (1,44) = 43.94, p < 0.0001), the main effect of factor treatment was significant; Factor age×factor treatment (F (1,44) = 35.07, p < 0.0001), the interaction effect between two factors was significant. Post-hoc Bonferroni showed that comparing to the control at PND60, prenatal poly I:C treatment decreased entries into the novel arm (p < 0.001) and no difference was found at PND40 (p > 0.05) (Fig. 2.2 A). Two-way ANOVA showed that factor age effect F (1,44) = 6.764, p < 0.05, the main effect of factor age was significant; Factor treatment effect F (1,44) = 4.586, p < 0.05, the main effect of factor treatment was significant; Factor age × factor treatment (F (1,44) = 2.775, p > 0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that comparing to the control at PND60, prenatal poly I:C treatment decreased time spent in the novel arm (p < 0.05) and no difference was found at PND40 (p>0.05) (Fig. 2.2 B).

Step-through passive avoidance test
In the passive avoidance memory test, two-way ANOVA showed that factor age effect (F (1,44) = 0.3814, p > 0.05), the main effect of factor age was insignificant; Factor treatment effect (F (1,44) = 0.5311, p > 0.05), the main effect of factor treatment was insignificant; Factor age × factor treatment (F (1,44) = 0.01102, p > 0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that there was no significant difference between the two groups at PND40 or PND60 on the first day (T1) (p > 0.05) (Fig. Supplement  1A). Two-way ANOVA showed that factor age effect (F (1,44) = 1.310, p > 0.05), the main effect of factor age was insignificant; Factor treatment effect (F (1,44) = 0.0008438, p > 0.05), the main effect of factor treatment was insignificant; Factor age × factor treatment (F (1,44) = 2.323, p > 0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that there was also no difference for latency time(T2 -T1) to re-enter the black box(p > 0.05) ( Fig. Supplement 1B).

Barnes Maze Test
In the Barnes maze test, two-way RM ANOVA showed that factor trial effect F (3,88) = 53.41, p < 0.0001, the main effect of factor trial was significant; Factor treatment effect (F (1,88) = 0.09667, p > 0.05), the main effect of factor treatment was insignificant; Factor trials × factor treatment (F (3,88) = 0.02535, p > 0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that there was no significant difference on the latency to find the target hole on each trail during the four days between the two groups at PND40 (p > 0.05) (Fig. 2.2 C). Two-way RM ANOVA showed that factor trial effect (F (3,88) = 5.457, p < 0.01), the main effect of factor trial was significant; Factor treatment effect (F (1,88) = 3.326, p > 0.05), the main effect of factor treatment was insignificant; Factor trials × factor treatment (F (3,88) = 0.2221, p > 0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that no significant difference was bound for the number of errors made in finding the target hole for the Barnes maze at PND40 (p > 0.05) (Fig. 2.2 D). Two-way RM ANOVA showed that factor trial effect (F (3,88) = 24.60, p < 0.0001), the main effect of factor trial was significant; Factor treatment effect (F (1,88) = 24.04, p < 0.0001), the main effect of factor treatment was significant; Factor trials × factor treatment (F (3,88) = 4.628, p < 0.01), the interaction effect between two factors was significant. Post-hoc Bonferroni showed that the latency to find the target hole was significantly longer in the poly I:C treated offspring at PND60 compared to the saline treated at the 1 st day (p < 0.001) and the 2nd day (p < 0.01) (Fig. 2.2 E). Two-way RM ANOVA showed that factor trial effect (F (3,88) = 21.51, p < 0.0001), the main effect of factor trial was significant; Factor treatment effect (F (1,88) = 1.031, p>0.05), the main effect of factor treatment was insignificant; Factor trials × factor treatment (F (3,88) = 0.5450, p>0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that no significant difference was bound for the number of errors at PND60 (p > 0.05) (Fig. 2.2 D). These results suggest that the effect of GD9 poly I:C treatment on spatial exploration, spatial recognition memory, and working memory, is progressive impairment from adolescence to adulthood.

Pre-pulse inhibition (PPI) test
Impairments in hearing sensory gating were detected by the PPI test, two-way RM ANOVA showed that factor trial effect (F (2,66) = 2.436, p>0.05), the main effect of factor trial was insignificant; Factor treatment effect (F (1,66) = 25.35, p < 0.001), the main effect of factor treatment was significant; Factor trials × factor treatment (F (2,66) = 0.01305, p>0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that compared to the control groups, prenatal poly I:C treatment significantly reduced % PPI for each pre-pulse intensity (75, 80, and 85 dB) in offspring at PND60 (p < 0.05) (Fig. 2.2 H). Two-way RM ANOVA showed that factor trial effect (F (2,66) = 1.821, p>0.05), the main effect of factor trial was insignificant; Factor treatment effect (F (1,66) = 5.001, p < 0.05), the main effect of factor treatment was significant; Factor trials × factor treatment (F (2,66) = 0.08256, p>0.05), the interaction effect between two factors was insignificant. Post-hoc Bonferroni showed that no difference for any pre-pulse intensity was observed at PND40 (p>0.05) (Fig. 2.2 G). Together, these behavioral results demonstrate that MIA induced by a single injection of 10 mg/kg poly I:C on GD9 results in schizophrenialike behavior and other cognitive impairments in offspring.

Prenatal immune activation induces age-related alterations of NMDA receptors and PSD95 in the PFC and HIC
NMDA receptors play a role in synaptic formation and function. In order to clarify maturation-dependent alterations in NMDA receptors and the postsynaptic protein PSD95, we analyzed levels of three NMDA receptor subunits (NR1, NR2A, and NR2B) and PSD95 in the HIC and PFC at different stages (PND21, PND45, and PND65).
At PND21, the expression of NMDA receptors in MIA offspring showed a significant decrease when compared to saline-treated dams in both the PFC and HIC (p < 0.05) (Fig. 3). However, no difference in PSD-95 was observed. The transcription levels of all genes in the PFC and HIC were detected and consistent with the protein levels (p < 0.05) (Fig. 3). The transcription and protein levels of NMDA subunits in both brain regions at PND45 showed significant intra-group differences but no inter-group differences, except for NR2A levels that were elevated in the PFC(p < 0.05) (Fig. 4). The poly I:C-treated young adult offspring showed significantly higher levels of all NMDA subunits and PSD95 in the PFC (Fig. 5), although only NR2B and PSD95 had higher transcriptional levels (p < 0.05) (Fig. 5). Only NR2A was upregulated in the HIC as indicated by mRNA and protein (p < 0.05) (Fig. 5).

Immunofluorescence studies
NMDA receptors are composed of NR1 and NR2 subunits which bind glutamate and exert a regulatory control over NR1. The NR1 subunit is encoded by one gene to produce several splice variants. The antibody used in this work was selected because it interacts with all of known splice variants. The change in NR1 expression, therefore, reflects a real overall change in NMDA receptor number or function. Two-way ANOVA showed that factor age (F (2,12) = 5.475, p < 0.05), the main effect of factor age was significant; Factor treatment (F (1,12) = 2.031, p>0.05), the main effect of factor treatment was insignificant; Factor age × factor treatment (F (2,12) = 6.713, p < 0.05), the interaction effect between two factors was significant. Post-hoc Bonferroni showed that pups at PND21 exposed to poly I:C all exhibited reduced NR1 immunoreactivity in the HIC when compared to the control group (p < 0.01). No significant difference was found in NMDA receptor expression in CA1 during puberty (p>0.05) and young adult (p>0.05) (Fig. 6A,C). Two-way ANOVA showed that factor age (F (2,12) = 8.567, p < 0.01), the main effect of factor age was significant; Factor treatment (F (1,12) = 18.56, p < 0.001), the main effect of factor treatment was significant; Factor age × factor treatment (F (2,12) = 14.83, p < 0.001), the interaction effect between two factors was significant. Post-hoc Bonferroni showed that the level of NR1 immunoreactivity in the PFC of poly I:C-treated young adult offspring was markedly higher than in the control groups (p < 0.001), and there was no difference at PND21 and PND45 (p > 0.05) (Fig. 6B,D). Since NR1-like immunoreactivity was predominately distributed in the somato-dentritic compartment of the CA1 neurons, changes of NR1 in CA1 were consistent with the overall observed changes. These data suggest that prenatal immune activation does have long time effects on NMDA receptor expression.

Correlational analysis of NMDA receptor expression and behavioral impairment
Children diagnosed with anti-NMDA receptor encephalitis have deficits in attention and visual motor function that involves executive functional impairments. To explore whether NMDA receptor expression correlated with behavioral impairment in this model of prenatal immune challenge, correlations were sought between behavioral factors and the expression of NMDA receptors at PND65. Cognitive indices were correlated with protein levels of NMDA receptor, using first-order partial correlations controlling for prenatal treatment (control or Poly I:C). Briefly, protein levels of NR2B in the PFC areas of interest showed a significant negative correlation with entries into the open arms (r = −0.610, df = 11, n = 14, p < 0.05; Fig. 7A). Protein levels of NR2A in the PFC correlated with cognitive performance in the Y-maze spatial novelty preference test (r = −0.574, df = 11, n = 14, p < 0.05; Fig. 7B). These results demonstrate NMDA receptors expression at PND65 to correlate with behavioral impairment.

Discussion
The "two-hit" neurodevelopmental model supports the concept that pre-perinatal events result in later abnormal neural processes during adolescence or young adulthood when psychotic symptoms typically emerge. This maturational delay suggests that study of molecular abnormalities at an early stage, when behavioral abnormalities have not yet occurred, may provide insight into the etiology and prevention of disease. Studies of this type that examine molecular and behavioral changes are important but few in number. The data herein provide the first line of experimental evidence demonstrating that early prenatal exposure to immune activation results in age-related abnormalities in behavior and NMDA receptors. Further, correlations between NMDA receptors and behavioral impairment, including anxiety, learning and  memory were demonstrated. These results suggest that NMDA receptor abnormalities precede schizophrenia-like behaviors, with the prospect that NMDA receptors are a potential target to prevent disease development during the asymptomatic period of schizophrenia. Such information may be useful for the development of preventive and/or therapeutic strategies for schizophrenia. A growing body of evidence sheds light on the neurodevelopmental nature of schizophrenia with symptoms typically emerging during late adolescence or young adulthood. We compared behavior at the presymptomatic adolescence period and the full symptomatic period of adulthood in the poly I:C MIA rat model and found that the behavioral abnormalities in MIA offspring were progressive, with more serious symptoms observed in adults rather than adolescents. This is consistent with other derived from poly I:C-induced gestational infection animal models and clinical features of schizophrenia [30]. Adolescent offspring of MIA display a more active state as indicated by increased spontaneous movement, the absence of anxiety, impaired working memory, and PPI deficits when compared to controls. Adult offspring of MIA showed decreased spontaneous movement, increased anxiety, impaired working memory, suppression of exploratory behavior, and sensorimotor gating impairment.
The abnormal behaviors observed in adult offspring of MIA were consistent with previous reports. For example, previous studies have reported that maternal LPS treatment increased anxiety levels in adult offspring [31,32]. Furthermore, several animal studies have indicated conflicting effects of MIA on abnormal behaviors. Shirin Babri et al., have reported that no significant change in anxiety behavior was found in adult c57BL/6 MIA offspring [31]. As indicated earlier in the introduction, the dose and timing of immunogen administration during gestation in animal strains can affect the effects of maternal immune activation on behavior [31].
The first episode of psychosis usually occurs in late adolescence or early adulthood, but premorbid impairments occur much earlier. One primary causal factor for schizophrenia is dysfunction of glutamate receptors [33][34][35][36]. In the PFC and HIC, MIA-rats showed abnormal NMDA receptor expression or distribution prior to and with the development of behavioral deficits from weaning to adolescence and to adulthood, reflecting an early impairment during the asymptomatic period. Studies suggest that activation of the immune system can interfere with NMDA receptors function during brain development. However, dynamic changes in NMDA receptors over time are unknown. To our knowledge, this is the first detailed study to observe NMDA receptor changes over time in a poly I:C model. Consistent with our data on the period of weaning, Yuko Fujita et al. have reported NMDA receptor hypofunction in P28 rat offspring after MIA. Supplementation with D-serine, an endogenous co-agonist of the NMDA receptor, prevented the onset of cognitive deficits in adult offspring after MIA. Tasnim Rahman et al. have studied the effects of immune activation on NMDA receptors in adult rat offspring and reported widespread increases in NR2A binding, no change in NR2B binding for poly I:C off ;spring. The increase in NR2A was also observed in adult rats exposed postnatally (at 2 weeks of age) to poly I:C [37] or the bacterial mimic LPS [38], suggesting the long-term regulatory effect of immune activation on NR2A. The variable findings for NR2B in adult rat offspring of MIA could be due to subtle diff ;erences in the timing of immunogen administration during gestation, or the methods of detection. The upregulated NMDA receptor subunits may indicate over activation of NMDA receptors, perhaps leading to glutamatergic excito-toxicity in MIA offspring at adulthood. Further detailed studies on how maternal poly I:C exposure induces NMDA receptors abnormalities are needed.
Numerous studies have shown the contribution of NMDA receptor disorders to behavioral impairment. Direct evidence comes from anti-Nmethyl-D-aspartate receptor encephalitis, which is characterized by behavior problems including; inattention, temper tantrums, hyperactivity, or irritability [39][40][41]. NR2B has been shown in animals to impact cognitive deficits, as well as anxiety and depressive-like behavior [42]. Abnormalities of the major subunit of the NMDA receptor in this poly I:C rat model likely explain some of the behavioral abnormalities, since correlations between NMDA subunits and behavioral impairments were found. Lehner M. et al. found that in anxiety rats exposed to chronic restraint stress, the level of NR2B in hippocampal and cortical regions was changed [43]. Researches also suggested that NR1, NR2A, NR2B are involved in learning and memory [44]. Consistent with previous animal results, we found the expression of NR2B in HIC and PFC to be correlated with anxiety, with learning and memory correlated with the levels of NR2A. Further research is required to determine whether NMDA receptor abnormalities contribute to the behavioral dysfunction observed in adult MIA rat offspring. The postsynaptic scaffolding protein PSD95 tethers the NMDA receptors to other proteins, forming an NMDA receptor signaling complex that transduces glutamate signals in postsynaptic cells. Our findings indicated that PSD95 was increased in the PFC of adult rats, which may result from invalid synaptic pruning and reorganization [45] following synaptic injury in MIA offspring. Given the physiological role of PSD95 in the central nervous system and its functional relationship with NMDA receptors, up-regulated PSD95 may indicate over activation of NMDA receptors, perhaps leading to glutamatergic excite toxicity in adult MIA offspring. It's interesting that increased PSD95 in MIA offspring was not observed till adulthood, which coincided with more serious behavioral impairment. Previous studies showed that viral-like prenatal immune activation in mice impairs the expression of PSD95 in the HIC [46,47]. Expression within the PFC was not assessed. The reason for PSD95 differences in the HIC may be due to differences in methods and time of maternal immune activation. In any case, prenatal poly I:C exposure in rats clearly induces PSD95 dysfunction in offspring.
One limitation of this study is that the association between changes in NMDA receptors and behavioral abnormalities were not investigated. Additional work is required to delineate the contribution of NMDA receptors to changes in behavioral abnormalities. This may be accomplished by pharmacological and/or genetic approaches that would analyze whether distinct NMDA receptor abnormalities can serve as clinical therapeutic targets.
In conclusion, these data suggest that early prenatal immune activation induces age-related behavioral and NMDA receptor alterations in the rat, demonstrating that this prenatal immune activation model matches the neurodevelopmental hypothesis of schizophrenia. Herein prenatal immune activation changed neurodevelopmental trajectories that may interact with maturational processes precipitating the full spectrum of NMDA receptor abnormalities throughout life. The alterations in NMDA receptors occurred much earlier than schizophrenia-like symptoms in the MIA offspring, which suggests that early effects are more important than those in adults to prevent schizophrenia. Furthermore, correlations between NMDA receptors and behavioral impairments suggest NMDA receptors as potential therapeutic targets for schizophrenia.

Declarations of interest
The authors declare no competing interest.

Author's contributions
LXL and WQL designed the study and wrote the protocol. KKH and XS established the animal model of MIA and wrote the manuscript. BBL and YQC helped in sample preparation and interpretation of the study. YFY, MLS, and MS undertook the statistical analysis and helped with molecular biology techniques. All authors contributed to and have approved the final manuscript.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding authors upon reasonable request.

Consent for publication
Not applicable.

Ethics approval and consent to participate
All experimental protocols and procedures were approved by the Animal Care and Use Committee of the Henan Key Lab of Biological Psychiatry (Xinxiang, China).