Maternal Immune Activation by Polyinosinic-Polycytidylic Acid Exposure Causes Cerebral Cortical Dysgenesis through Dysregulated Cell Cycle Kinetics of Neural Stem/Progenitor Cells

Maternal immune activation reportedly causes dysregulation of the cell cycle in stem cells and impairment of higher cortical function in rodents. Furthermore, in humans, maternal immune activation during the first to second trimester of pregnancy is strongly correlated with increased incidence of autism spectrum disorder in the offspring. Here, we show that in utero exposure to polyinosinic-polycytidylic acid (poly (I:C)) in mice during the early phase of neuronogenesis increases the probability of differentiation (quiescent fraction [Q fraction]) of neural stem/progenitor cells (NSPCs) without change in the length of cell cycle. This abnormal increase in the Q fraction is assumed to reduce the peak population size of NSPCs, resulting in the thinning of the neocortex in offspring because of the reduced production of neurons. Furthermore, the neocortex of poly (I:C)-exposed mice does not exhibit a layer-specific reduction in radial thickness, possibly because of increased apoptosis caused by poly (I:C) exposure during all stages of cortical development. These results suggest that maternal immune activation by poly (I:C) exposure may affect neocortical histogenesis by altering the cell cycle kinetics of NSPCs. In addition, the timing and amount of poly (I:C) exposure during pregnancy may have profound effects on cerebral cortical histogenesis.


Introduction
necessarily infect the fetus directly; rather, maternal viral infection increases the maternal level of inflammatory cytokines, which then pass through the placenta to reach developing embryos.
To investigate the effects of maternal infection on fetal neural development, immune activation induced by maternal exposure to a double-stranded RNA, polyinosinic-polycytidylic acid (poly [I:C]), is a wellestablished mouse model for maternal RNA-viral infection [2,3]. Poly (I:C) directly binds to maternal toll-like receptor 3 and increases the maternal expression levels of cytokines, including interleukin (IL)-6, IL-1β, and tumor necrosis factor (TNF)-α [4]. Using this model, researchers have shown that mice born from mothers exposed to poly (I:C) exhibit abnormalities in social interaction/behavior and repetitive activities that resemble the symptoms of ASD in human children [5]. In addition, cerebral cortical dysgenesis in mice born from mothers exposed to poly (I:C) has been reported. Specifically, (1) the production of projection neurons in the superficial layer of the cortex is delayed [6], (2) the number of interneurons in the somatosensory area of the cortex is reduced [7], and (3) the transition probability of Tbr2positive neural stem/progenitor cells (NSPCs) from pax6-positive NSPCs is reduced in the offspring of poly (I:C)-exposed mothers [8]. These observations are supported by reports describing laminar disorganization and agenesis of specific laminae in the cerebral cortex in autistic children [9].
We previously reported that in utero exposure to environmental factors, such as dioxin [10] and the antiepileptic drug valproic acid [11], affected the cell cycle kinetics of NSPCs and led to cerebral cortical dysgenesis; both of these reports were based on analyses performed using a mathematical model of cerebral cortical histogenesis [12,13]. These results indicate that neuronal production is indeed affected by disruptions in cell cycle kinetics caused by dioxin and valproic acid exposure in utero.
Considering the above background, we hypothesized that maternal immune activation caused by poly (I:C) exposure may affect the cell cycle kinetics of NSPCs, which would lead to an abnormal probability of differentiation (the fraction of daughter cells that leave the cell cycle; quiescent fraction [Q fraction]) and/or length of the cell cycle. Additionally, abnormal cell cycle kinetics might lead to cerebral cortical dysgenesis. Thus, in this analysis, we used a mouse model to expose pregnant mothers to poly (I:C) and analyzed the cell cycle kinetics of embryonic NSPCs and cerebral cortical histogenesis in the offspring.

Materials and Methods
Animal Maintenance and Poly (I:C) Administration Pregnant CD-1 mice (Sankyo Labo Service, Tokyo, Japan) were maintained under a 12-h light-dark schedule. Toll-like receptor ligand-tested polyinosinic-polycytidylic acid sodium salt (poly (I: C), P1530, Sigma-Aldrich, St. Louis, MO, Lot # 088M4116V and # 80809) was dissolved in diethyl pyrocarbonate (DEPC)-treated phosphate-buffered saline (PBS) at a concentration of 2 mg/mL. The solution was heated for 10 min at 55°C and cooled for 10 min at room temperature. A single dose of poly (I:C) solution, or DEPC-treated PBS (0.01 mL/g body weight [bw]) as control, was injected peritoneally to pregnant CD-1 mice on embryonic day (E) 12. The total dose of poly (I:C) administered was 20 mg/kg bw. All the experimental procedures were in full compliance with the institutional guidelines of the Laboratory Animal Center, Keio University School of Medicine.

Measurement of Inflammatory Cytokines
Whole blood was obtained from the submandibular vein of pregnant mice 3 h after the injection of poly (I:C) solution or DEPC-treated PBS peritoneally on E12. Sera were collected and processed by Milliplex MAP Mouse TH17 Magnetic Bead Panel (MTH17MAG-47K-06, Merck Millipore, Burlington, MA, USA) according to the manufacturer's protocol. We analyzed IL-1β, TNFα, interferon (IFN) γ, IL-6, IL-10, and IL-17A with the panel. The body weight of the pregnant mice was measured daily until the birth of the offspring. Nine poly (I:C)-treated and three control mice were examined for inflammatory cytokines.
Tissue Processing for Histological Analysis Postnatal mice were deeply anesthetized by medetomidine (0.3 mg/kg bw), midazolam (4.0 mg/kg bw), and butorphanol (5.0 mg/kg bw). The anesthetized mice were perfused transcardially with 4% phosphate-buffered paraformaldehyde (Sigma-Aldrich); the whole brain was immersion fixed with 4% paraformaldehyde for overnight at 4°C and embedded in paraffin. The analysis was conducted using serial coronal sections with 4 μm thickness, which included the dorsomedial primary somatosensory area of neocortices (field 1) [14].
As for the analysis in embryos, whole heads were immersion fixed with 4% paraformaldehyde overnight at 4°C and embedded in paraffin. Serial coronal sections of forebrains/telencephala were cut at a 4 μm thickness, including the dorsomedial regions of the cerebral walls that correspond to future field 1 [15].

Cumulative Labeling Analysis
Pregnant mice were administered 5-iodo-2'-deoxyuridine (IdU, Sigma-Aldrich, 50 μg/g bw) and/or 5-bromo-2'-deoxyuridine (BrdU, Sigma-Aldrich, 50 μg/g bw) by intraperitoneal injection starting from 9:00 a.m. on E14. The mice were sacrificed at 2-, 4.5-, 6.5-, and 12.5-h after initial IdU or BrdU administration. In the 4.5-and 6.5-h experiments, we initially administered IdU and readministered BrdU 2 h after IdU injection. In the 6.5-and 12.5-h experiments, we administered BrdU every 3 h after initial BrdU injection to maintain its concentration in embryos. The sections from embryonic forebrain were processed for immunohistochemistry using anti-BrdU antibody (1:50, AbD Serotec, Oxford, UK) as a primary antibody overnight at 4°C and Alexa Fluor 555 anti-rat IgG antibody (1:1,000, Thermo Fisher Scientific, Waltham, MA, USA) as a secondary antibody for 1.5 h at room temperature. In the 4.5-and 6.5-h experiments, we double immunostained with anti-IdU/BrdU (1:12, Becton Dickinson, Franklin Lakes, NJ, USA) and anti-BrdU antibodies (1:50, AbD Serotec) overnight at 4°C and Alexa Fluor 488 anti-mouse IgG antibody (1:1,000, Thermo Fischer Scientific) and Alexa Fluor 555 anti-rat IgG antibody (1:1,000, Thermo Fischer Scientific) as secondary antibodies for 1.5 h at room temperature. The sections were counterstained with bisBenzimide H33342 (1:300 of 1% solution, Sigma-Aldrich) for 1.5 h at room temperature. The labeling index (LI), the ratio of the number of IdU/BrdU labeled to the total number of nuclei, was used as the analytical index. Analysis was conducted at the dorsomedial cerebral wall in a sector measuring 100 μm in length in the coronal plane. Nuclei of endothelial cells were not counted. The average LI was calculated for each bin across a series of five to nine brains obtained from the embryos of three to four litters.

Q Fraction Analysis
Cells in the Q fraction were identified using two S-phase tracers, IdU and BrdU, as previously described [10,16,17]. Briefly, pregnant CD-1 mice that were exposed to poly (I:C) or PBS on E12 were administered IdU (50 μg/g bw) by intraperitoneal injection at 7:00 a.m. on E14. At 9:00 a.m., the animals were divided into two groups: (1) one group in which BrdU (50 μg/g bw) injection was administered every 3 h and then sacrificed 12.5 h later (Q experiment) and (2) a second group in which a single BrdU injection was administered, followed by sacrifice of the animals 12.5 h afterward (P + Q experiment). Both sets of forebrains were fixed in 4% phosphate-buffered paraformaldehyde and embedded in paraffin, sectioned serially into 4-µm sections, and double immunostained with anti-IdU/BrdU (Becton Dickinson) and anti-BrdU antibodies (AbD Serotec). The number of IdU-positive/ BrdU-negative nuclei was counted in the dorsomedial ventricular zone (VZ) of four to six nonadjacent sections from nine brains obtained from the embryos of three litters.

Number and Laminar Fate of the Q Cells in the Postnatal Neocortices
IdU and BrdU were administered to the pregnant mice likewise the Q experiment on E14, and the coronal sections of field 1 of neocortices on postnatal day (P) 21 were made [18]. In addition to the antibodies used in the Q fraction analysis, anti-Cux1 antibody (1:50, Santa Cruz Biotechnology, Dallas, TX, USA, 4°C, overnight) was applied as a marker of layers II-IV of the neocortices. The number of IdU-positive/BrdU-negative neurons, i.e., the "E14born" Q cells, was counted. Fifteen brains from five litters were examined for the poly (I:C)-treated mice and six brains from two litters for controls. For each brain, four nonconsecutive sections were analyzed.

Terminal Deoxynucleotidyl Transferase Mediated dUTP Nick End Labeling Staining
Embryonic forebrains on E14 were immersion fixed in 4% phosphate-buffered paraformaldehyde and embedded in paraffin, sectioned coronally at 4 μm, and the sections were processed for terminal deoxynucleotidyl transferase mediated dUTP nick end labeling (TUNEL) analysis. TUNEL-positive nuclei were detected by a commercially available kit (ApopTag Plus Peroxidase In Situ Apoptosis Detection Kit, Millipore). The number of TUNEL-positive nuclei was counted in the cerebral wall of four to six nonadjacent sections from five to six embryonic forebrains that were obtained from three litters.

Statistical Analysis
Statistical significance was evaluated by two-tailed Student's t tests in which p values < 0.05 were regarded as statistically significant. Additionally, analysis using a linear mixed-effects model was conducted when indicated.

Poly (I:C) Exposure Decreased Maternal Body Weight and Increased IL-6 Expression
The body temperature of the mothers exposed to poly (I: C) on E12 decreased by 1.19°C from the baseline (−1.19 ± 0.33°C vs. −0.15 ± 0.19°C in the controls, p = 0.013, Fig. 1a). The body weight of the mothers exposed to poly (I:C) on E12 was decreased by 4.7% (49.3 ± 0.7 g vs. 47.0 ± 0.9 g in the controls, p = 0.0009), compared with the values in the controls on E13 (Fig. 1b). However, the body weight recovered and did not decrease again after E14 (Fig. 1b). When the expression levels of inflammatory cytokines (IFNγ, IL-1β, IL-6, IL-10, IL-17A, and TNFα) in sera from both groups of mothers were analyzed 3 h after poly (I:C) injection on E12, IL-1β (p = 0.028), IL-6 (p = 0.0005), IL-10 (p = 0.015), and TNFα (p = 0.005) levels were statistically increased in the poly (I:C)-exposed mothers' sera, compared with that of the controls (Fig. 1c). A negative correlation was seen between the expression level of IL-6 and the change in body weight of the poly (I:C)-exposed mothers from E12 to E13 (Spearman r = −0.8095, p = 0.022, Fig. 1d).

Maternal Poly (I:C) Exposure Did Not Alter the Cell Cycle Length of Embryonic NSPCs
We examined poly (I:C)-treated embryonic forebrains on E14, the time point at which glutamatergic neurons in layer IV are produced [18]. Histologically, the dorsomedial cerebral wall, which is the future primary somatosensory neocortex corresponding to area 1, was normal in the poly (I:C)-treated embryos (Fig. 2a). The S-phase zone in the dorsomedial cerebral wall, where the accumulation of NSPC nuclei is observed during S phase of the cell cycle, was located between 70 and 80 μm from the lateral ventricular border on E14 in poly (I:C)-treated mice, similar to the findings in normal control mice (Fig. 2b) poly (I:C)-treated and control mice (Fig. 2b). Since the LI for 2-, 4.5-, and 6.5-h BrdU exposure did not show any difference between the poly (I:C)-treated and control embryos (Fig. 2b), we concluded that the total cell cycle length of the NSPCs in the forebrain of the poly (I:C)treated animals was not significantly altered, compared with that in the controls.

Maternal Poly (I:C) Exposure Increased the Q Fraction of Embryonic NSPCs
We next identified NSPCs in the Q fraction and the P fraction (the fraction of daughter cells that remain proliferative, P = 1.0 -Q) on E14 using two S-phase tracers, IdU and BrdU (Fig. 3a, b). As mentioned previously, we conducted a Q fraction analysis based on the assumption that the total cell cycle length minus the length of the S phase (Tc -Ts) was 12.5 h for both the poly (I:C)-treated and control groups. In the dorsomedial cerebral wall of the poly (I:C)-treated mice, the IdUpositive nuclei were located in the outer margin of the S-phase zone in both the P + Q and Q experiments (Fig. 3b). The distribution patterns of the P + Q and Q cells were identical between the poly (I:C)-treated and control animals (Fig. 3c). However, an increase in the number of Q cells was observed in the poly (I:C)-treated animals, with an estimated Q fraction of 0.55; this represented a 12.5% increase, compared with the value in the controls (Table 1). Taken together, we concluded that the probability of leaving the cell cycle of the NSPCs in the forebrain of the poly (I:C)-treated animals was increased, compared with that in the controls.

Maternal Poly (I:C) Exposure Increased the Number of Q Cells and Affected the Migration Pattern of Q Cells in the Neocortex, as Assessed on P21
We observed an increased number of Q cells that were born on E14 in the primary somatosensory cortex of the poly (I:C)-treated mice on P21, compared with that in the controls (9.45 ± 0.46 vs. 7.91 ± 0.77 cells, p = 0.047; poly [I:C] group, 15 brains from 5 litters; control group, 6 brains from 2 litters) (Fig. 5a, b). Interestingly, the distribution pattern of the Q cells within the same area of cortex was altered by maternal poly (I:C) exposure in utero (Fig. 5a, c). In the control mice, most of the Q cells that were born on E14 were distributed in layer IV of the cortex. However, the Q cells of the poly (I:C)-treated offspring were distributed more broadly within layers III to V (Fig. 5c). These results indicated that not only the number of Q cells but also the mechanism that governs the migration pattern was affected by maternal poly (I:C) exposure in utero.

Maternal Poly (I:C) Exposure Increased the Number of Apoptotic Nuclei in the Embryonic Forebrain
We identified the number of TUNEL-positive nuclei in the cerebral wall on E14 to investigate the effect of maternal poly (I:C) exposure on apoptosis. The number of TUNEL-positive nuclei in the cerebral wall on E14 was increased by maternal poly (I:C) exposure compared to those of controls (5.45 ± 1.31 vs. 1.93 ± 0.06 cells per counting area in controls, p = 0.016; poly [I:C] group, 6 brains from 3 litters; control group, 5 brains from 3 litters) (Fig. 5d, e).

Discussion
In the present study, we described the alteration in the cell cycle kinetics of the NSPCs by maternal immune activation as detected by cumulative labeling and Q fraction analyses in utero. In addition, as far as we know, there are no reports regarding the effect of maternal immune activation on neuronal migration that partially causes nonspecific layer thinning of the neocortex. In combination with increased apoptosis in the embryonic cerebral wall, we concluded that impaired cell cycle kinetics of the NSPCs and migration characteristics of  daughter neurons, presumably due to disturbance of molecular mechanisms within the NSPCs by exposure to inflammatory cytokines, may result in abnormal cortical histogenesis in offspring.

Timing of Administration and Amount of Poly (I:C) Exposure in Utero
Epidemiological data from human subjects have indicated that maternal immune activation during the first to second trimester of pregnancy is strongly correlated with an increased incidence of ASD in offspring [19]. Thus, in the present analysis, we exposed maternal mice to poly (I:C) on E12, which corresponds to the early phase of neuronogenesis, and examined the cell cycle kinetics of NSPCs on E14 [13]. In addition, the effect of maternal poly (I:C) exposure on offspring behavior is known to differ among mouse models. Since C57BL/6 mice have a lower sensitivity to poly (I:C) exposure, compared with CD-1 mice, a previous study applied 10-20 mg/kg bw of poly (I:C) to C57BL/6 mice to induce abnormal behavior that was similar to that detected in CD-1 mice administered 5 mg/kg bw of poly (I:C) [20]. However, in our preliminary analysis, the administration of 5-10 mg/kg bw of poly (I:C) to pregnant CD-1 mice did not induce histological abnormality in the offspring. Considering this background, we applied 20 mg/kg bw of poly (I:C) to investigate neocortical histogenesis in the following experiments.

Mechanism Underlying Increased Q Fraction in Response to Maternal Immune Activation
The increased Q fraction that is observed in NSPCs on E14 in response to poly (I:C) exposure is compatible with a previous report showing that in utero exposure to lipopolysaccharide on E12 increased the production of young neurons by promoting NSPCs to exit the cell cycle [21]. Since poly (I:C) is reported to induce p27 Kip1 protein expression in human dendritic cells, maternal immune activation might increase cell cycle regulatory proteins that promote cell cycle exit [22]. Although the mechanism of maternal immune activation by lipopolysaccharide exposure is unlikely to be identical to that of poly (I:C), we concluded that poly (I:C) exposure in utero did indeed have a persistent abnormal effect on cell cycle kinetics on E14, 2 days after poly (I:C) exposure. The increased Q fraction on E14 seems to decrease the peak number of NSPCs, reducing the total number of projection neurons that are produced; we speculated that this phenomenon led to a decreased thickness of the neocortex. The increased number of neocortical Q cells on P21, which were born on E14, also partially supports this observation.
However, previous reports have controversial results regarding the changes in the Q fraction of NSPCs. Maternal poly (I:C) exposure on E9 was reported to delay the phenotype switch of neurons from deeper layer neurons to superficial layer neurons, suggesting that the Q fraction seems to be abnormally decreased during the early phase of neuronogenesis [6]. In another analysis, maternal poly (I:C) exposure on E16, which corresponds to a later phase of neuronogenesis, increased the number of Pax6-positive NSPCs and decreased the number of Tbr2-positive NSPCs on E18 [8]. This observation suggests that the reduction in the Q fraction seems to be followed by an increase in the number of Pax6-positive NSPCs. In these reports, however, the doses of poly (I:C) administered to the C57BL/6 mice were not the same as those used in the present report, in which 20 mg/kg bw of poly (I:C) was administered to CD-1 mice. Thus, these factors might play a substantial role in this difference in the alteration pattern of the Q fraction of the NSPCs.
The Reduction in the Size of the Telencephalon in Response to Maternal Immune Activation We observed reduced width of the telencephalon by 4.0% (9.80 ± 0.101 vs. 10.21 ± 0.105 mm) in the poly (I:C)-treated mice, compared with those of the controls (Fig. 4b). The difference in the width between both groups, i.e., 0.410 mm, indicates 205 μm reduction for each right and left sides of the telencephalon. As the cortical thickness of the dorsal area of neocortex (primary somatosensory cortex) was reduced by 9.8% (677.3 ± 9.11 vs. 751.1 ± 13.13 μm) in the poly (I:C)-treated mice, compared with those of the controls (Fig. 4c, d), we consider that the reduction in width in the poly (I:C)-treated mice may be a combination of reduced thickness of the lateral side of cortex and reduced volumes of the inner structure of the telencephalon (globus pallidus, caudate putamen, thalamus, etc.). Although we did not measure the thickness of lateral side of the cortex and the volume of the basal ganglia in the present analysis, we speculate that the reduced width is compatible with the reduced cortical thickness.
Non-Layer-Specific Thinning of the Cerebral Cortex in Response to Poly (I:C) Exposure In the present analysis, thinning of the neocortex was observed in both the superficial and deep layers, which was not compatible with our previous observations. The abnormal increase in the Q fraction during the early phase of neurogenesis decreased the thickness of the deeper layer and not the superficial layer in two independent experiments using NSPC-specific p27 Kip1 overexpression and in utero dioxin exposure analyses [10,17,23]. We speculated that the mixed effects of NSPC exposure to several cytokines might play a substantial role in nonspecific layer thinning in the neocortex. A series of cytokine exposure experiments indicated that some cytokines promote apoptosis, while others protect against apoptosis [24]. As we analyzed the number of TUNEL-positive nuclei in the cerebral wall of poly (I:C)exposed embryos on E14, poly (I:C) exposure in utero indeed increased the number of apoptotic nuclei, suggesting that the overall effect of poly (I:C) exposure in our experimental model was the promotion of the apoptosis pathway.
In addition, we detected the abnormal migration pattern of Q cells within the neocortex in poly (I:C)-exposed mice. This phenomenon might also result in nonspecific layer thinning of the neocortex. In this context, p27 Kip1 is also known to regulate RhoA activation and to modulate cell migration [25]. Although we do not have conclusive data regarding the expression profiles of cell cycle regulatory proteins in the present analysis, the expression level of p27 Kip1 protein might affect the rate of migration of Q cells within the neocortex.

Severity and Timing of Inflammation Induced by Poly (I:C) in Utero
In the present study, we applied a single high dose of poly (I:C) to pregnant CD-1 mice because of the abovementioned preliminary result. However, in this scenario, it may not be suitable to reproduce mild to moderate inflammation that persists for a certain period of time during pregnancy. The in utero exposure of 5 mg/kg bw of poly (I: C) for 5 days (either E10-E14 or E14-E18) reportedly results in an increased thickness of the cerebral cortex in offspring [26]. Another point that should be discussed is the induction level of cytokines. A previous report has indicated that low and high levels of exposure to TNFα increase the rate of survival and increase the apoptosis of NSPCs, respectively [24]. This result suggests that the expression level of even a single cytokine might have an opposite effect on the cellular fate of NSPCs.
In summary, poly (I:C) exposure of 20 mg/kg bw during the early phase of neuronogenesis increased the Q fraction of NSPCs and induced apoptosis in the embryonic cerebral wall on E14. These abnormalities finally resulted in a reduced thickness of the neocortex in a nonlayer-specific manner. We speculated that the timing and the amount of poly (I:C) exposure during pregnancy may result in inconsistent effects on cerebral cortical histogenesis, and these factors might affect the cell cycle kinetics of NSPCs in different manners.