A CCR5 antagonist, maraviroc, alleviates neural circuit dysfunction and behavioral disorders induced by prenatal valproate exposure

All animal procedures were performed in accordance with the Fundamental Guidelines for Proper Conduct of Animal Experiments and Related Activities in Academic Research Institutions under the Jurisdiction of the Ministry of Education, Culture, Sports, Science and Technology, Japan. The Animal Care and Use Committee of Hiroshima University approved the experimental protocols (No. C18-23-4). Pregnant ICR mice were obtained from Japan SLC (Shizuoka, Japan) and were maintained in a temperature-controlled animal facility on a 12-h light–dark cycle (light on from 8:00 a.m. to 8:00 p.m.). Mice were allowed ad libitum access to diet and water. Sodium valproate (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in saline. Pregnant female mice received a single dose of 800 mg/kg sodium valproate by gavage on day 11.5 after conception. Minocycline (Wako Pure Chemical, Osaka, Japan; 100 mg/kg/day) was dissolved in drinking water at the required concentration, which was determined based on each dam's body weight and drinking volume, and given to the dams from 1 day old (P1) to 21 days old (P21) via lactation. Maraviroc (GlaxoSmithKline, London, UK; 80 mg/kg/day) was dissolved in drinking water, sonicated, warmed and administered to the dams from P1 to P21 via lactation. The administration schedule is shown in Fig. 1. We used simple randomization for the randomization method [26]. Four to five mice were housed in the same cage from P22. The brain tissues were perfused with phosphate-buffered saline under isoflurane anesthesia and then isolated. No sample calculation was performed and that sample size was determined by our previous studies and other reports [25, 27‐29]. Any animal was not excluded based on the statistics or died during experiments. One hundred thirty-two dams and 322 male pups were used in this study. N values and details of litters are included in figure legends as appropriate.

Eye opening was sequentially observed after birth and scored as 0.5 for one eye open and 1 for both eyes open to evaluate development. Behavioral tests were performed by well-trained investigators who were blinded to the experimental groups. Data analyses were also performed by technical staff independently. Behavioral tests were performed in the order of open field test, Y-maze test, social affiliation test and marble burying test from the age of 6 weeks. All behavioral tests were performed from 8:00 am to 12:00 am. In the open field test, the mice were released into an open field arena (90 × 90 × 50 cm). The arena was equipped with a camera mounted overhead. The mice were monitored for 90 min, and the distance moved was calculated by DIPP-Motion V/2D software (DITECT, Tokyo, Japan).

The Y-maze test was performed using an apparatus with three arms (20 cm long × 10 cm wide × 20 cm high) at 120° angles. The mice were placed in the distal end of an arm and allowed to explore the maze for 8 min. The percentage of alternations (entries into an arm that differed from the arm entered in the previous two entries) was calculated with the following formula: (alternations/(arms entries – 2)) × 100 [30].

A social affiliation test was performed in a rectangular apparatus (20 cm long × 40 cm wide × 40 cm high). Empty wire containment cups were placed on the right and left sides 8 cm from the wall, and the mice were habituated to the apparatus for 5 min. A control mouse (stranger) was placed in a wire containment cup that was located on one of the sides of the apparatus, and then the mice were placed in the apparatus for 10 min. A video camera mounted above the apparatus recorded the movements of the mice for analysis. The amount of time spent around each cage (empty cage or stranger-containing cage) was measured using DIPP-Motion V/2D software [31].

The marble burying test was performed in a mouse cage (22 cm long × 32 cm wide × 15 cm high), in which 20 glass marbles (diameter 1.2 cm) were spaced evenly on a 2-cm-deep layer of shaved wood bedding. The mice were placed in the cage and left undisturbed for 30 min. After the test, the number of marbles buried (covered more than two-thirds with bedding) was counted [32].

IHC was performed according to our previous report [33]. Briefly, the brains were fixed with 4% buffered paraformaldehyde and cryoprotected in 30% sucrose. Brains were frozen, and 50-μm-thick floating sections were prepared using a Cryostat (CM3050 S; Leica Biosystems, Nussloch, Germany). The sections were blocked and permeabilized with PBS containing 10% normal goat serum (Sigma-Aldrich) and 0.3% Triton-X 100 for 1 h at room temperature. The sections were incubated with primary antibody for 3 h at room temperature, followed by secondary antibody for 1 h at room temperature in the dark. The antibodies used are listed in Additional file 1: Table S1. The sections were mounted on a glass slide with DAPI-Fluoromount-G (Southern Biotech, Birmingham, AL, USA). Images were processed using the Zen image acquisition software package (Carl Zeiss, Oberkochen, Germany) and ImageJ software. The soma area was evaluated using images of Iba1 staining, and the amoeboid score was calculated by the following formula: (soma area/entire Iba1-stained area) × 100 (%). The Iba1- and CD68-costained areas were analyzed with ImageJ software (National Institutes of Health, Bethesda, MD, USA). At least 100 cells were measured for calculating the amoeboid score and CD68-stained area.

Minocycline levels in the serum and brain were measured as described in a previous report with slight modifications [34]. Hippocampal homogenates were prepared in phosphate-buffered saline. Fifty microlitres of each homogenate was mixed with 20 μL of 0.5 M dipotassium hydrogen phosphate aqueous solution and 250 μL of ethyl acetate. The mixture was vortexed and centrifuged, and the resulting upper layer (ethyl acetate layer) was isolated. Fifty microlitres of 20 mM hydrochloric acid was added to the ethyl acetate solution. The mixture was vortexed and centrifuged, and 20 μL of the lower layer (water layer) was injected onto an InertSustain Cyano column (4.0 × 250 mm, 5 µm) (GL Science, Tokyo, Japan). The mobile phase consisted of methanol and 20 mM perchloric acid/4 mM triethylamine in water (20:80, v/v; pH approx. 2). The flow rate was 1 mL/min, the column temperature was 25 °C, and UV detection was performed at 350 nm. Solutions of minocycline (Wako Pure Chemical) were used as standards.

The maraviroc concentration was determined as described in a previous report with slight modifications [35]. Equal amounts of serum or hippocampal homogenates were mixed with acetonitrile. After the mixtures were vortexed and centrifuged, 50 μL of supernatant was injected onto an InertSustain Cyano column (4.0 × 250 mm, 5 µm) (GL Science). The mobile phase consisted of acetonitrile and 20 mM potassium dihydrogenphosphate in water, and the pH was adjusted to 2.5 by phosphoric acid (40:60, v/v). The flow rate was 0.6 mL/min, the column temperature was 25 °C, and UV detection was performed at 210 nm. Solutions of known concentrations of maraviroc (Cayman Chemical Company, Ann Arbor, MI, USA) were used as standards.

VSD imaging was performed using the same techniques as those described in previous reports [36, 37]. Briefly, hippocampal slices (350 µm) were prepared and transferred onto a fine-mesh membrane filter (Omni Pore membrane filter, JHWP01300; Millipore) held in place by a thin Plexiglas ring (inner diameter, 11 mm; outer diameter, 15 mm; thickness, 1–2 mm). The slices were stained for 25 min with 100 µL of VSD solution [0.2 mM di-4-ANEPPS (Thermo Fisher Scientific, Waltham, MA, USA) in 2.5% ethanol, 0.13% Cremaphor EL (Sigma-Aldrich), 1.17% distilled water, 48.1% FBS, and 48.1% artificial cerebrospinal fluid (ACSF)]. The slices were used for experiments after at least a 1-h incubation at room temperature following VSD washout.

The Plexiglas ring supporting each slice (see above) was placed in an immersion-type recording chamber [38]. Custom laboratory-designed epifluorescence optics consisting of a custom-made objective lens (Brainvision 6×, water immersion) and a Leica Microsystems MZ-APO (f = 55 mm × 1.0) projection lens were used to view the slices during the experiments. Excitation light was provided by a high-power stabilized LED light (LEX-2G; Brainvision Co Ltd, Tokyo, Japan) projected through an excitation filter (λ = 530 ± 10 nm) and reflected onto the hippocampal slice by a dichroic mirror (λ = 575 nm). Emission fluorescence from the slice passed through an emission filter (λ > 590 nm) and was projected onto a C-MOS imager (MiCAM-02; Brainvision). The intensity of fluorescence emitted by the slice prior to stimulation was averaged and used as the reference intensity (F0). The fractional change in fluorescence [ΔF(t) = F(t)-F0] was normalized to F0 (ΔF/F0), and this value was used as the optical signal. The optical signals referred to below represent signals filtered in spatial and temporal dimensions with a digital Gaussian kernel of 5 × 5 × 3 (horizontal × vertical × temporal; σ ≈ 1). At a wavelength of 610 nm, VSD fluorescence decreased in response to depolarization of the cell membrane. Electrical stimulations (40 V, bipolar 200 µs) were applied with constant voltage pulses (ESTM-8, Brainvision Co. Ltd.) through a glass microcapillary tube (5 µm inner diameter; filled with ACSF) placed either on Schaffer collateral afferents in the CA3/CA1 border of CA1, on the granule cell layer to stimulate the mossy fiber pathway, or in the molecular layer of the upper blade of the dentate gyrus (DG). Stimulation was applied at an interval of at least 30 s.

RNA was prepared from the hippocampi of 5-day-old mice exposed to vehicle or VPA using a ReliaPrep RNA Miniprep System (Promega, Madison, WI, USA). RNA isolated from 5 male pups from 2 dams in each group was analyzed by CAGE-seq. RNA quality was assessed with a Bioanalyzer (Agilent, Santa Clara, CA, USA) to ensure that the RNA integrity number (RIN) was over 8.0 and that the A260/A280 and 260/230 ratios were over 2.0. CAGE library preparation, sequencing, mapping, and gene expression analysis were performed by DNAFORM (Yokohama, Kanagawa, Japan). First-strand cDNA was transcribed to the 5′ end of capped RNAs and attached to CAGE barcode tags, and these tags were sequenced using the NextSeq 500 system (Illumina, San Diego, CA, USA) and mapped to the mouse mm9 genomes using the BWA software program after discarding ribosomal RNAs. Over 20 million reads were mapped to the murine genome sequence for each sample. The data were analyzed, and the expression ratio was also calculated as the log (base 2) ratio using the RECLU pipeline. Raw data were registered in the NCBI GEO database (accession no. GSE180564).

mRNA levels were determined according to the protocol described in our previous report [39]. Briefly, total RNA was extracted from the hippocampus using a High Pure RNA Isolation Kit (Roche Diagnostics K.K., Tokyo, Japan). Single-stranded cDNA was synthesized from 1 μg of total RNA according to the ReverTra Ace protocol (Toyobo, Osaka, Japan) with a random primer (9-mer; Takara Bio, Shiga, Japan). Real-time PCR was performed using a CFX Connect real-time PCR system (Bio–Rad Laboratories, Hercules, CA, USA) with TB Green Premix Ex Ta II (TaKaRa Bio, Shiga, Japan). The primer sequences are presented in Additional file 1: Table S2. mRNA levels were normalized to the level of the housekeeping gene β-actin, and the values of the treated samples were divided by those of the untreated samples to calculate the relative mRNA levels.

Freshly isolated hippocampi were lysed in radioimmunoprecipitation assay (RIPA) buffer containing cOmplete EDTA-free Protease Inhibitor Cocktail (Roche). The hippocampal CCL3 concentration was determined using the Mouse CCL3/MIP-1 Alpha Quantikine ELISA Kit (R&D Systems, Inc., Minneapolis, MN, USA) according to the manufacturers’ instructions.

CD11b-positive cells were isolated according to our previously described method [40]. Cells or tissues were lysed with RIPA buffer (25 mM Tris–HCl (pH 7.6), 150 mM NaCl, 1% Nonidet P-40, 1% sodium deoxycholate, and 0.1% SDS). Equal amounts of protein were loaded, separated via SDS-PAGE and transferred onto polyvinylidene difluoride membranes. The blocked membranes were incubated with the primary and secondary antibodies listed in Additional file 1: Table S1. Then, the membranes were visualized using peroxide substrates (SuperSignal West Dura, Thermo Fisher Scientific).

All data were analyzed using GraphPad Prism 9 (GraphPad Software, San Diego, CA, USA). Student's t test, one-way ANOVA with Dunnett's corrected multiple comparison tests, two-way ANOVA [VPA × Age (Fig. 2A), VPA × time points (Fig. 2B), VPA × Mino (Figs. 3B–E, G, 4B, D, F and 6A, B) or VPA × Mar (Fig. 7B–E, G)] with Tukey's corrected multiple comparison tests were used to determine significant differences between the means of two or more independent groups of animals. The F value calculated from ANOVA is described in each figure legend, the p value is indicated in each figure, and significance was considered when p values were less than 0.05. Error bars were calculated using SD.

Pregnant mice were orally administered VPA on embryonic day 11.5 (E11.5) (the protocol is shown in Fig. 1), and the VPA concentration in the maternal serum increased to approximately 73 μg/mL, which is almost equivalent to the therapeutic range in humans (from 50 to 100 μg/mL), 3 h after treatment (Additional file 1: Fig. S1). VPA was detected in the fetal brain 3 h but not 24 h after administration (Additional file 1: Fig. S1), indicating that VPA could be transiently translocated into the fetal brain and then excreted immediately. Neural development, as evaluated by eye opening, was delayed in VPA-treated mice (Fig. 2A). Although fetal VPA exposure did not affect exploratory behavior or locomotor activity (Fig. 2B), VPA decreased the alternation, which is an index of spatial working memory defects, and decreased the amount of time spent exploring around mice, indicating abnormal social affiliation (Fig. 2C and D). VPA-treated mice also showed an increase in the number of buried marbles (Fig. 2E), suggesting that prenatal VPA exposure elicited repetitive behaviors. The number of microglia in the hippocampal CA1, CA3 and DG regions was similar between postnatal day 10 (P10) vehicle-treated mice and VPA-treated mice (Fig. 2F and G). However, the microglial soma area was significantly increased in the CA1 region by prenatal VPA exposure, and CD68 protein levels were also increased in VPA-treated mice (Fig. 2F, H and I). Flow cytometry showed that peripheral macrophages did not invade the hippocampus in P10 VPA-treated mice (data not shown). Prenatal VPA exposure showed no change in the expression of glial fibrillary acidic protein (GFPA) and S100β in the hippocampus of P10 mice, suggesting that VPA does not affect astrocytic activity (Additional file 1: Fig. S2). Thus, these data clearly showed that prenatal exposure to VPA can activate microglia in the hippocampal CA1 region during the early postnatal periods.

There was no change in microglial number, soma size or CD68 expression in the hippocampal CA1 region of VPA-treated mice at the age of 6 weeks (Additional file 1: Fig. S3A–D). However, when basal neuronal responses upon electrical stimulation were measured, prenatal VPA treatment significantly increased excitatory activity in the hippocampal CA1 region at 6 weeks of age (Additional file 1: Fig. S3E and F). Therefore, these changes in excitation/inhibition balance in the hippocampus might contribute to abnormal behavior elicited by VPA treatment.

Minocycline (7-dimethylamino-6-dimethyl-6-deoxytetracycline), a second-generation semisynthetic tetracycline analog, is widely used to prevent the activation of microglia [41]. When minocycline was administered from P1 to P10 via lactation, the minocycline concentrations in the maternal serum and the hippocampi of the pups at P10 were 1.20 ± 0.10 μg/mL and 0.56 ± 0.05 μg/g tissue, respectively, indicating that minocycline was accumulated in the pup brain when administered via lactation. Iba1/CD68 staining showed that microglial activation, such as soma enlargement and increased CD68 expression in the CA1 region, induced by VPA was largely suppressed by minocycline administration (Fig. 3A–D). Importantly, behavioral disorders induced by VPA, such as defects in social affiliation and repetitive behaviors, were also alleviated by minocycline treatment via lactation from P1 to P21 (Fig. 3F and G). Unfortunately, since treatment with minocycline alone exacerbated special working memory (Fig. 3E), we could not evaluate the effects of activated microglia on special working memory. Minocycline has been reported to have multiple functions in the brain, including the action into the endocannabinoid system and the inhibition of dopamine release [42, 43]. In addition, our unpublished data showed that hippocampal levels of brain-derived neurotrophic factor were largely decreased by minocycline treatment according to our experimental protocol described here (data not shown). Therefore, these actions of minocycline in addition to microglial suppression might be involved in the deterioration of special working memory.

Basal neuronal responses in the three major synaptic connections in the hippocampus (CA3-CA1, Schaffer collateral afferent; DG-CA3, mossy fibers; EC-DG, perforant pathway) were measured upon electrical stimulation, and it was found that prenatal VPA exposure significantly increased excitatory activity, especially upon stimulation of Schaffer collaterals, in P10 mice and that this excitation–inhibition imbalance was clearly suppressed by minocycline treatment (Fig. 4A, B). VPA treatment tended to elicit an excitation–inhibition imbalance upon mossy and perforant stimulation, but the effect was not significant (Fig. 4C–F). Taken together, these results indicate that microglial activation during development elicited by fetal VPA exposure can cause neural circuit dysfunction and behavioral disorders after development. As shown in Additional file 1: Fig. S2E and F, excitation–inhibition imbalance in the CA1 region of VPA-treated mice remained until the age of 6 weeks, when behavioral abnormalities were detected. Therefore, overexcitation in the hippocampus during the developmental stage might cause abnormal post-growth behaviors.

VPA inhibits histone deacetylase in addition to GABA transaminase [44]. Therefore, gene expression in the hippocampi of P5 vehicle- and VPA-treated mice was analyzed by CAGE-seq. Genes that were upregulated 4 times or downregulated 4 times in the VPA-treated group compared with the vehicle-treated group and with p values lower than 0.1 were extracted, and genes reportedly expressed in the brain were selected (Additional file 1: Table S3). The expression of the 7 selected genes (lgals7, wfdc2, lbp, fgfr1, cntfr, tlr13 and ccl3) was confirmed by real-time PCR. CCL3 mRNA levels showed significant increases in the hippocampi of VPA-treated mice compared with those of vehicle-treated mice (Fig. 5); thus, we focused on the role of CCL3 in prenatal VPA exposure-induced developmental defects.

The increase in CCL3 mRNA expression in the hippocampus of P10 mice induced by prenatal VPA treatment was significantly suppressed by minocycline administration, and the increase in CCL3 protein expression induced by VPA was also attenuated by minocycline (Fig. 6A and B). Next, CD11b-positive cells, mainly microglia in the brain, were immunoprecipitated from the P10 hippocampi, and CCL3 protein expression was evaluated by western blotting. CCL3 expression in CD11b-positive cells was significantly increased by prenatal VPA exposure (Fig. 6C and D), suggesting that CCL3 expression is upregulated in hippocampal microglia at least after birth. CCL3 mRNA levels in the cortex and amygdala were not affected by VPA (Fig. 6E). No change in the expression of CCR1 and CCR5, which are receptors for CCL3, was observed between vehicle- and VPA-treated mice (Fig. 6F). In addition, CCL3 expression was not increased in the hippocampi of P1 mice (Fig. 6G).

The maternal inflammatory state can influence neonates through a process known as maternal immune activation (MIA) [45]; however, VPA administration did not cause an increase in the serum levels of proinflammatory cytokines such as IL-1β and IL-6 (Additional file 1: Fig. S4). Therefore, MIA is unlikely to contribute to developmental microglial activation in this VPA exposure model. CCL3 was reported to be upregulated epigenetically in macrophages, and the upstream region of the CCL3 gene has a fundamental role in the upregulation [46]. Thus, to identify the mechanism by which CCL3 expression is upregulated in the hippocampus by VPA treatment, chromatin immunoprecipitation (ChIP) was performed. DNA was extracted from CD11b-positive cells in the hippocampus and precipitated with anti-histone H3 and H4 antibodies and anti-acetyl-histone H3 and H4 antibodies. Real-time PCR was performed using a primer designed upstream of the CCL3 gene, according to the report by Kiguchi et al. [46]. More of the upstream region of the CCL3 gene was detected in precipitated DNA with anti-acetyl-histone H3 prepared from VPA-treated mice than vehicle-treated mice (Additional file 1: Fig. S5A), suggesting that histone H3 upstream of the CCL3 gene can be acetylated in the hippocampal microglia of VPA-treated mice. Western blotting confirmed that histone H3 acetylation in the hippocampus increased in the VPA-treated group compared with the vehicle-treated group (Additional file 1: Fig. S5B and C). Taken together, these results suggest that CCL3 expression may be upregulated via an epigenetic mechanism that includes histone H3 hyperacetylation, at least in part.

CCR1 and CCR5 are the main receptors for CCL3. Among these, CCR5 is known to be largely distributed in the brains of several species, including humans and mice [47]. Thus, we attempted to alleviate neural circuit dysfunction and abnormal behaviors induced by fetal VPA exposure with a CCR5 antagonist, maraviroc (UK-427857). When maraviroc was administered via lactation from P1 to P10, the maraviroc concentrations in the maternal serum and hippocampi of the pups were 92.7 ± 10.0 ng/mL and 331 ± 53 ng/g tissue, respectively, indicating that maraviroc was able to enter the brains of the pups. Maraviroc did not affect microglial activity in the hippocampus of mice prenatally exposed to VPA (data not shown). Maraviroc clearly suppressed the increase in excitability induced by prenatal VPA exposure in the hippocampal CA1 region of P10 mice (Fig. 7A–D). Maraviroc administration also suppressed VPA-induced post-developmental defects in spatial working memory and social affiliation and repetitive behaviors (Fig. 7E–G). Therefore, maraviroc can recover post-developmental behavioral disorders by alleviating the excitation–inhibition imbalance during development elicited by prenatal VPA treatment.

Several VPA administration protocols are used to study prenatal VPA-induced developmental disorders [8, 18]. Thus, CCL3 expression in the hippocampus was assessed after VPA administration rather than at the dose of 800 mg/kg VPA orally on E11.5. Three applications of 300 mg/kg VPA orally at E12.5, 13.5 and 14.5 to dams significantly increased hippocampal CCL3 levels at P10 (Fig. 8A). Intraperitoneal treatment with 500 mg/kg VPA at E12.5 in dams also increased CCL3 expression (Fig. 8A). Therefore, CCL3 upregulation in the developing hippocampus is considered one of the general mechanisms by which prenatal exposure to VPA induces post-developmental behavioral disorders.