Introduction

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized mainly through social impairments, restricted and repetitive behavior and, interests. So far, no pharmacological therapy has been found to treat the main symptoms of this disorder effectively, and most drugs target co-morbidities, even though an increasing portion of the population is being diagnosed with ASD (Association 2013; Lord et al. 2018). Several genetic and environmental factors have been mentioned as risk factors for autism. An environmental factor that increases the risk of autism is maternal exposure to valproate (VPA) during the first trimester of pregnancy (Al Sagheer et al. 2018; Lord et al. 2020). The researchers developed an animal model for autism induction based on this and observed clear signs of autism in offspring following VPA injection in pregnant rats ((Schneider et al. 2008b). Neurological abnormalities during development are responsible for the particular behaviors associated with autism. Several brain areas are associated with autism, including the hippocampus, prefrontal cortex, cerebellum, and amygdala, but no conclusive structural abnormalities have yet been observed (Lord et al. 2020; Mizuno et al. 2019).

Amygdala dysfunction has been considered in some neurodevelopmental disorders. The amygdala is involved in long-term memory and recognition of the emotional meaning of stimuli and is where memory and emotion come together. Pleasant emotional learning and fearful emotional learning, is linked to the amygdala (Ruggieri 2014). Based on animal research, it has been suggested that the amygdala, in connection with the orbitofrontal cortex (OFC) and the superior temporal gyrus (STG), forms the neural basis of social intelligence, and the whole of this neural network is called the “social brain” (Zalla and Sperduti 2013). The amygdala plays an essential role in processing social-emotional stimuli as well as in anxiety. Given the function of the amygdala, it is not surprising that much emphasis has been placed on its involvement in the etiopathology of autism. It has been seen that the structural development of the amygdala is altered during infancy and in children with autism (Bauman and Kemper 1985; Herrington et al. 2017; Li et al. 2019; (Truitt et al. 2007a).

Inhibitory (GABAergic) parvalbumin-positive interneurons (PVI) are a large subpopulation of neurons in the amygdala, constituting 50% of basolateral amygdala interneurons (Sah et al. 2003). Inhibitory interneurons, especially the GABAergic type, play an essential role in information processing throughout the brain. They provide balance against excitatory neurotransmitters by creating hyperpolarization in the postsynaptic neuron. Regulation of information transmission and neuronal function depends on this inhibitory regulation (Ko et al. 2015; Marín 2012; Rudolph and Möhler 2014). Since PVIs need a lot of energy for their function, they are subjected to oxidative stress and loss of mitochondrial membrane potential. So, they are very vulnerable to stressors. In general, a change in this inhibitory system causes various psychiatric disorders such as anxiety and autism (Kann et al. 2014; (Ruden et al. 2021a; Whittaker et al. 2011). Considering the above information, in the present study we focused on involvement of amygdala in development of valproate autism model and anxiety-related behaviors with emphasis on parvalbumin interneurons and mitochondrial dysfunction.

Materials and methods

Experimental design

Adult Wistar rats (Purchased from Pasteur’s Institute, Tehran, Iran), housed in ventilated cages at the physiology department animal facility (Tehran University of Medical Sciences), were allowed free access to food and water on a 12 h light/dark cycle and were kept at a temperature of 23 °C. Female and male rats were used for mating. In the aftermath of mating, as evidenced by the presence of vaginal plugs, males were retrieved from their cages. Pregnant rats received a single dose of either valproate (VPA, 600 mg/kg) or saline at gestational day 12.5 (E12.5), intraperitoneally (Nicolini et al. 2015). Then, pregnant rats were left undisturbed until delivery. At postnatal day 23 (P23), offspring were weaned, separated from their mothers, grouped and kept in the animal room for one week to be adapted for next behavioral tasks (Lee et al. 2018; Saré et al. 2021). Due to the higher incidence of ASD in males compared to females, as well as studies that reported that prenatal VPA exposure failed to induce ASDs in female rodents (Nicolini and Fahnestock 2018; (Schneider et al. 2008a), only male offspring were used in this study. Male pups were divided in two groups of 10 randomly.1) Control group; animals in this group are offspring of mothers who received saline during pregnancy; 2) Valproate group; animals in this group are offspring of mothers who received VPA during pregnancy. The average weight of male offspring at P30 was 98 g in control group and 82 g in valproate group. Following behavioral analysis, animals in both experimental groups were sacrificed. Brains were removed for further investigation, as described below. The experimental method shown in Fig. 1. This study followed the guidelines laid out in the National Institutes of Health’s Guide for the Care and Use of Laboratory Animals (NIH). The study procedure was accepted by Tehran University of Medical Science’s ethics committee (IR.TUMS.MEDICINE.REC.1400.564).

Fig. 1
figure 1

Experimental design. All rats were maternally administered with saline or valproate (VPA, 600 mg/kg) at embryonic day 12.5. After weaning, male offspring were divided into two groups. Following habituation, behavioral tests were conducted

Full size image

Behavioral tasks

All behavioral tests were conducted between 8:00 am and 2:00 pm in an acoustic and dim-light room specialized for behavioral tests and recorded manually. The behavioral test apparatuses are made of black and matte plexiglass. After each animal test, the apparatus was cleaned with 70% ethanol and allowed to evaporate. We started the behavioral tests with an open field test and then continued with elevated zero maze and U-Shaped 2 Choice Field, respectively.

Open field test

To assess anxiety-like behavior, at P31 and P32, open field test (60 × 60 × 40 cm) was used. Gridlines in the floor of the apparatus divided the area into 16 equal smaller squares (15 * 15 cm squares). 12 squares along the walls and 4 squares in the center. The subject animal was placed at the center of the open field box and allowed to explore the box for 5 min. The number of entrances to the center zone and time spent in the center of the open field were measure. Animals with more anxiety spend less time in the center zone and show less entry in the center area than rats without anxiety.

Elevated zero maze

For further analysis of anxiety behavior, we placed the animals in an elevated zero maze (at P33 and P34). Elevated zero maze is an annular shape maze (with a diameter of 110 cm and a width of 10 cm) composed of 4 consecutive open and closed areas (two open areas and two walled and enclosed parts). The apparatus is placed at a height of 70 cm from the floor. During the five-minute test, the rat was placed in an open area of the elevated zero maze and allowed to explore freely. The number of entrances and time spent in each open and closed area were recorded.

Social interaction task

To assess sociability and social novelty in animals, the U-Shaped 2 Choice Field was used. The U-Shaped 2 Choice Field is an open rectangular box (60 × 60 cm and 40 cm in height). One of the walls of this box is divided into two parts by a wall placed from the middle of the wall towards the center of the box. Two grid cages were placed within the quadrants that are closed, forming a U-Shaped apparatus to give subjects with two choices. U-Shaped 2 Choice Field conducted in three phases, (1) habituation: the subject rat was placed in the apparatus with empty grid cages and explored for 10 min. At the end of this phase, animal returned to its home cage, (2) sociability and social interaction task: the social target rat (stranger 1) was placed in one of the grid cages. The subject was placed in the middle space of the apparatus and allowed to explore both sides for 10 min, and (3) social novelty recognition task: At this phase, earlier stranger (stranger 1) was placed in its grid cage, and a new social target animal (stranger 2) was placed in the other grid cage. The subject was placed in the middle space of the apparatus and allowed to explore both sides for 10 min. The time spent on each side of the apparatus was recorded. Social targets were allowed to acclimatize in the grid cage for 1 min before testing began.

Immunoblotting

To isolate amygdala, rat brain matrix was used and 2 mm coronal sections were prepared for isolation of amygdala on ice under a stereomicroscope and according to stereotaxic coordinates. For western blotting, punched-out amygdala tissue was lysed with RIPA buffer. After centrifuging (for 20 min. at 4 °C at 14,000 rpm), the lysates were separated. The protein content was quantified using the Bradford protein kit under the recommendations of the manufacturer. The lysates and 2X Laemmli sample buffer were mixed together with an equal volume. After boiling for 5 min, lysates (20 mg) were then transferred to an immune-blot PVDF membrane and subjected to SDS-PAGE (Bio-Rad Laboratories, USA). The membranes were then blocked with BSA (5%) (Sigma Aldrich, USA) in Tween 20 (0.1%) for 1 h. Then, the membranes were incubated with anti- AMPK (AMP-activated protein kinase) (Abcam, USA), anti- SIRT1 (Sirtuin 1) (Abcam, USA), anti- PGC1α (Peroxisome proliferator-activated receptor-gamma coactivator 1 alpha) (Abcam, USA) and anti-beta actin-loading control antibodies (Abcam, USA) for 1 h at room temperature. After three TBST washes, membranes were treated with a secondary goat anti-rabbit IgG (HRP) (Abcam, USA) antibody. After that, the membranes underwent a 2-minute enhanced chemiluminescence (ECL) incubation. Protein expression was normalized to β-actin, and the resultant values were then compared between groups (Babaei et al. 2018).

Biochemical parameters of the amygdala

Following the behavioral test, a combination of ketamine and xylazine (80 and 10 mg/kg, respectively; i.p.) was used to deeply anesthetize the rats (n = 7 from each experimental group), and amygdala tissue was isolated from the whole brain and homogenized in cold lysis buffer with protease inhibitors present. Afterward, the supernatant was analyzed to determine reactive oxygen species (ROS), mitochondrial membrane potential (MMP), and caspase-3 activity.

Determination of ROS level

The amount of ROS was measured by fluorimetry method using the fluorescent probe 2′,7′-dichlorofluorescein diacetate (DCF-DA) (Kiazist, Iran). This substance (DCF-DA) penetrates into cells and is then de-esterified by intracellular esterase enzymes and is trapped inside the cell. After being reduced by free oxygen species, it shows fluorescence properties. This was read at a wavelength of 525 nm in a plate reader after excitation at 488 nm.

Determination of MMP

Due to the potential difference between the inner and outer mitochondrial membrane, rhodamine cationic dye accumulates in the mitochondrial matrix. Fluorescent rhodamine 123 dye (a cationic dye) was used for the measurement of mitochondrial membrane potential as a specific probe, and the fluorescence was recorded by a fluorescent spectrophotometer (490 nm excitation and 520 nm emission) (Baracca et al. 2003; Keshavarz-Bahaghighat et al. 2018).

Activity of caspase-3

Caspase-3 activity was measured using a colorimetric method and according to the instructions of its kit from Kiazist, Iran. Working solution contained caspase buffer, dithiothreitol, and enzyme substrate. After addition of 50 µl of sample to 55.5 µl of working reagent and its incubation at 37˚C for 120 min, absorbance was obtained at 405 nm.

Immunohistochemistry

The brains of anesthetized rats were perfusion-fixed, and sliced into coronal Sect. (5 μm thick) using a microtome (Leica, USA). For PV-neuron quantification in the amygdala, sections were incubated with mouse monoclonal anti-PV antibody (1:100, Sigma, USA) overnight, followed by incubation with secondary antibody (HRP-conjugated, 1:100, Santa Cruz Biotechnology, Inc., USA). PV-positive immunoreactivity was developed using 3′3′-diaminobenzidine-tetrahydrochloride and mild counterstaining with Hematoxylin and IOD for the intensity of PV immunoreactivity was determined in the amygdala region.

Data Analysis

All the parameters were analyzed with an unpaired t-test except social behavior data that analyzed with two-way ANOVA and Sidak post hoc test. Data are expressed as mean ± Standard Error of the Mean (SEM) and analyzed using GraphPad Prism-8 software. For all analyses, a p value of < 0.05 was considered significant.

Results

Effects of prenatal exposure to valproate on anxiety and social behavior

As presented in Fig. 2, animals with exposure to valproate prenatally show more anxiety and less social communication willingness compared to the control. The open field test data showed that valproate animals made fewer entries into the center zone (t = 5.297, Mean for control = 5.6, Mean for valproate = 2.1, 95% C.I. for the difference=-4.88 to -2.11, p < 0.001) and spent fewer time in the center zone (t = 3.116, Mean for control = 13.5, Mean for valproate = 6.7, 95% C.I. for the difference=-11.38 to -2.21, p < 0.01) (Fig. 2b and c). In the elevated zero maze test, the valproate animals made less open arm entries (t = 4.352, Mean for control = 3.9, Mean for valproate = 1.7, 95% C.I. for the difference=-3.26 to -1.13, p < 0.001) and spent less time on the open arms (t = 3.392, Mean for control = 42, Mean for valproate = 16.9, 95% C.I. for the difference=-40.65 to -9.552, p < 0.01) relative to control animals, respectively (Fig. 2e and f). Sociability and social novelty were assessed in U-Shaped 2 Choice Field maze. There was a significant difference between groups as determined by statistical analysis for sociability (F (1, 36) = 69.39, Mean for control = 244.1, Mean for valproate = 230.4, 95% C.I. for the difference= -29.26 to 56.56, p < 0.001) and social novelty preference (F (1, 36) = 14.93, Mean for control = 250.3, Mean for valproate = 269.9, 95% C.I. for the difference= -97.35 to 58.15, p < 0.001). Valproate rats exhibited no sociability and social novelty preference at a significant level compared to control rats, as shown in Fig. 2h and i.

Fig. 2
figure 2

The effect of prenatal exposure to valproate on anxiety and social behavior. Anxiety like behavior was measured in open field test and elevated zero maze by accounting number of entrances in center zone in OFT (b) and open arms in EZM (e). Also, time spent in center zone (c) and open arms (f) were recorded in order to measure anxiety. Sociability behavior (h) and social novelty preference (i) were determined by evaluating the time spent in either side U-Shaped 2 Choice Field maze. Data are expressed as the mean ± SEM (n = 10). **p < 0.01, ***p < 0.001 vs. control group, CCC p < 0.001 in the same group vs. other side of maze

Full size image

The effect of prenatal exposure to valproate on AMPK/ SIRT1/PGC1α in the amygdala

We assessed protein levels of AMPK, SIRT1, and PGC1α by western blot analysis in the amygdala region. There was a significant difference between groups assessed on the levels of AMPK, SIRT1, and PGC1α proteins. Western blot analysis showed a significant decrease in AMPK (t = 3.534, Mean for control = 100, Mean for valproate = 69.72, 95% C.I. for the difference=-54.08 to -6.49, p < 0.05), SIRT1 (t = 11.42, Mean for control = 100, Mean for valproate = 53.82, 95% C.I. for the difference=-57.40 to -34.96, p < 0.001), and PGC1α (t = 2.790, Mean for control = 100, Mean for valproate = 68.92, 95% C.I. for the difference=-62.00 to -0.15, p < 0.05) protein levels in the valproate rats (Fig. 3) as compared to the control group.

Fig. 3
figure 3

The effect of prenatal exposure to valproate on AMPK/ SIRT1/PGC1α pathway proteins level in amygdala. The level of AMPK, SIRT1 and PGC1α proteins were measured in amygdala by western blot analysis. Data were reported as a percentage of control. Data are expressed as the mean ± SEM (n = 3). *p < 0.05, ***p < 0.001 vs. control group

Full size image

The effect of prenatal exposure to valproate on the ROS level of the amygdala

Amygdala tissue was evaluated for reactive oxygen species level. There was a significant difference between the two groups assessed on the levels of reactive oxygen species (Fig. 4a). In the amygdala of valproate-induced autistic animals, ROS production increased (t = 3.864, Mean for control = 100, Mean for valproate = 174.4, 95% C.I. for the difference = 32.45 to 116.4, p < 0.01) as compared with the control animals.

The effect of prenatal exposure to valproate on mitochondrial membrane potential in the amygdala

Figure 4b displays the fluorescent intensity for Rhodamine 123 (a fluorescent dye that accumulates in mitochondria depending on the pH). There was a significant difference between groups for the levels of mitochondrial membrane potential (MMP). The fluorescence intensity decreased significantly in the valproate group compared to the control group (t = 3.417, Mean for control = 100, Mean for valproate = 81.44, 95% C.I. for the difference= -30.40 to -6.72, p < 0.01).

The effect of prenatal exposure to valproate on apoptosis marker in the amygdala

Data of caspase-3 activity is shown in Fig. 4c. A significant difference was identified between the groups. The activity of caspase-3 elevated in the valproate group as compared to the control (t = 2.994, Mean for control = 0.08, Mean for valproate = 0.11, 95% C.I. for the difference = 0.007 to 0.04, p < 0.05).

Fig. 4
figure 4

The effect of prenatal exposure to valproate on biochemical parameters in amygdala. The level of reactive oxygen species (ROS) (a), mitochondrial membrane potential (MMP) (b) and activity of caspase-3 (c) were assessed in amygdala. Data are expressed as the mean ± SEM (n = 7). *p < 0.05, **p < 0.01 vs. control group

Full size image

The effect of prenatal exposure to valproate on parvalbumin immunoreactivity in the amygdala

The immunoreactivity of parvalbumin-positive interneurons in the amygdala area was assessed. Significant differences were observed between groups in the immunoreactivity of PVIs in the amygdala. Prenatal exposure to valproate caused a significant decrease in the immunoreactivity of PVIs as compared to the control group (t = 5.62, Mean for control = 134.5, Mean for valproate = 112.6, 95% C.I. for the difference=-32.68 to -11.07, p < 0.01) (Fig. 5a and b).

Fig. 5
figure 5

The effect of prenatal exposure to valproate on PV-positive interneuron quantification in amygdala. To quantify the parvalbumin positive interneurons, their immunoreactivity in amygdala were observed (a and b) (n = 3). Data are expressed as the mean ± SEM. **p < 0.01 vs. control group. Black arrows show parvalbumin-positive cells

Full size image

Discussion

Considering the amygdala’s functions and the major role of the PVIs in promoting brain function, we aimed to determine the change of PVIs in the amygdala in the VPA model of autism and investigate the underlying mechanism of that. As a prerequisite, we have first characterized this model at the behavioral level and reported major deficits in social activities and anxiety levels. Furthermore, a reduction in PVIs was found in the amygdala of autistic animals. Then, we noted that the possible mechanism involved in the loss of PVIs might be the activation of the AMPK/ SIRT1/PGC1α pathway following increased oxidative stress.

A growing body of scientific evidence demonstrates autistic-like behavior such as impaired social behavior, restricted interests and behaviors, and anxiety in the VPA model of autism (Kataoka et al. 2013; Kim et al. 2014). Our data showed, lack of sociability and social novelty in VPA-exposed rats (using the U-Shaped 2 Choice Field test). Also, anxiety-like behaviors were observed in open field test and elevated zero maze in VPA compared to control animals.

In this study, we emphasized on the amygdala, which is among the brain areas associated with autism spectrum disorder. Amygdala is a heterogeneous structure composed of sub-nuclei which differ in their input and output regions. Predicting and coordinating responses to sensory and social stimuli is the function of the basolateral amygdala (BLA) complex (lateral, basal, and accessory basal nuclei) that is connected to the prefrontal cortex (PFC), and the orbitofrontal cortex (OFC) (Seguin et al. 2021; (Truitt et al. 2007b). In rodents, activating BLA projections to the medial prefrontal cortex (mPFC) increases anxiety, while inhibiting these projections dampens anxiety (Felix-Ortiz et al. 2016). The study of Seguin et al. showed, increased BLA volumes in ASD children that it was associated with increased social skills deficiency (Seguin et al. 2021). It has also been shown that children with ASD suffer from anxiety, which is related to alterations to the structure of the amygdala (Hennessy et al. 2022).

VPA is a teratogen factor with the most severe effects, including defects in the neural tube (Qiu et al. 2020). It is well known that prenatal exposure to VPA causes developmental neurotoxicity in children’s central nervous systems (Ogawa et al. 2007; Ornoy et al. 2015). According to prospective and retrospective studies, pregnancy exposure to VPA is associated with a three-fold increased risk of dysmorphic features and decreased intrauterine development (Christensen et al. 2013; Wood et al. 2015). It has been reported that mothers exposed to VPA during the first trimester of pregnancy increase their baby’s chances of developing autism (Deckmann et al. 2018; Ornoy et al. 2015). Previous studies have reported abnormal development trends in rodent models exposed to high doses of VPA in utero (hundreds of mg/kg), including neural tube defects. Prenatal exposure to VPA has been associated with ASD-like symptoms in human and animal models (Al-Askar et al. 2017; Nicolini and Fahnestock 2018). Several pathophysiological mechanisms have been mentioned for the VPA effect on the neural tube during the fetal period, which it causes ASD, such as neurotoxicity, reduced neurogenesis, neuroinflammation, and oxidative stress (Taleb et al. 2021). Oxidative stress occurs when there is an inequity between the generation of reactive oxygen species (ROS) and the body’s ability to detoxify them. Consequently, excess oxidative stress leads to increased lipid peroxidation and DNA damage (Al-Gubory et al. 2010; Taleb et al. 2018). Due to the immature antioxidant system in the fetus, VPA may produce embryonic variations because the brain is more susceptible to ROS than other fetal organs (Qiu et al. 2020). VPA exposure in animals was associated with elevated expression of apoptotic markers in the neuroepithelium. These apoptotic markers are hypothesized to be responsible for neural tube defects and altered embryonic signaling pathways, which may be caused by ROS’s role in promoting apoptosis in offspring (Tung and Winn 2011). As such, oxidative stress within the brain can play a pathological role in autism spectrum disorders by affecting different signaling pathways (Li and Zhou 2016). The data of our study showed that the amount of ROS in the amygdala of VPA exposed rats increased compared to the control animals.

AMPK (AMP-activated protein kinase) plays an essential role in maintaining cell energy balance and regulates cell survival, metabolism, and cellular homeostasis. So, AMPK signaling could be associated with neurodegeneration and CNS pathology (Muraleedharan and Dasgupta 2022). Cell structure, polarity, division, and growth and development are also regulated by AMPK (Lee et al. 2007). A research study in mice has demonstrated that AMPK protects hippocampal neurons from metabolic, excitotoxic, and oxidative insults (Culmsee et al. 2001). As one of the downstream pathways, AMPK activates the SIRT1 (NAD+-dependent type III deacetylase sirtuin 1) by increasing the intracellular NAD+/NADH ratio. Activation of SIRT1 results in the deacetylation and modulation of the activity of the peroxisome proliferator-activated receptor-gamma coactivator 1 alpha (PGC-1α) (Cantó et al. 2009). The PGC-1α is a powerful ROS regulator and links mitochondrial biogenesis with the response to oxidative stress. PGC-1α reduction influenced several ROS detoxifying enzymes, including SOD, CAT, and Gpx, as well as cortical oxidative stress and antioxidant capacity(Christoforou et al. 2011; Cunningham et al. 2007). Interestingly, PGC-1α protein is positioned in interneurons in rodent brains by the second week after birth. This temporal pattern of expression causes PV (Parvalbumin) development. PGC-1α is most highly expressed in GABAergic neurons of the central nervous system, particularly in cortical PVIs (Cowell et al. 2007). In general, with the reduction of PGC-1α, the antioxidant capacity of the cell and the protective effect of this factor on the neuron decreases, and it may lead to the start of the apoptosis process (Cowell et al. 2009).

In this study, we observed that the amount of AMPK and SIRT1 proteins in the VPA group decreases, which probably decreases PGC-1α and leads to disruption in PV protein maturation. Parvalbumin protein (a calcium-binding protein) has a buffering role for calcium ions in PVIs. If the amount of parvalbumin protein in the cell decreases, toxic amounts of calcium accumulate in the mitochondria, and this causes an increase in ROS, a decrease in the mitochondrial membrane potential (MMP), and a disruption in the electron transport chain ((Ruden et al. 2021b). In this regard, the amount of ROS and MMP was measured, and the data indicate the accumulation of ROS and the decrease of MMP in the VPA group. As a result, cytochrome C is released and leads to the activation of apoptosis pathways, and an increase in the Casp3 apoptosis factor was seen. Damage to PVIs, which are also very vulnerable, causes a disturbance in the balance of excitation and inhibition in the nervous system. We observed that the PVIs in the amygdala altered in the VPA group. Several studies indicated that damage to or dysfunction of PVIs contributes to the pathophysiology of autism spectrum disorder (Gandal et al. 2012; Gogolla et al. 2009; (Ruden et al. 2021b). One of the limitations of this study was the lack of immunohistochemistry for AMPK, SIRT1 and PGC1α on amygdala sections which is strongly suggested for future studies in this field.

Conclusion

In conclusion, our findings provide clear evidence for increased anxiety and autistic-like behaviors concomitant with changes in the amygdala in the brains of rats exposed to valproate during fetal life. The data indicated mitochondrial dysfunction through increased ROS and decreased mitochondrial membrane potential that might be due to disturbed AMPK/SIRT1/PGC1α pathway activation. In the whole process, the apoptosis pathway is activated, which leads to the reduction of parvalbumin interneurons and ultimately causes dysfunction of the amygdala.