Background
Autism spectrum disorder (ASD) is a group of neurodevelopmental pathologies characterized by persistent deficits in social communication and restricted, repetitive behaviors and interests [
1]. Symptoms appear in early childhood and persist throughout adulthood. Although both genetic and environmental causes have been proposed and tested [
2], their etiology remains unknown. There is currently no consensus on the underlying neuropathology, and many brain structures have been claimed to play a relevant role in ASD symptoms, e.g., hippocampus [
3,
4], prefrontal cortex [
5], and cerebellum [
3,
6].
While great strides have been made in the treatment of ASD, its effectiveness varies greatly depending on the case, and the underlying mechanisms of partially successful therapies are unknown [
7]. In particular, clinical studies suggest that early social stimulation is the most effective treatment for autistic children, who show significant improvements in social behavior (e.g. [
8‐
10]). Early social stimulation could counteract genetic and/or environmental risk factors, helping children to interact better with others [
11].
There is extensive evidence on the effects of early environmental enrichment on brain development in animal models. Enrichment has been demonstrated to diminish the effects of genetic risk and injury on brain malfunction [
12,
13], and environmental enrichment can revert many of the ASD-related behaviors in rodent models of autism [
14,
15]. Our aim is to analyze the effects of early social stimulation (a specific component of environmental enrichment) on ASD-related behavior in a mouse model of ASD and to assess its consequences on brain function.
It has been extensively shown that prenatal exposure of mice to valproic acid (VPA) at gestational day 12.5 results in reduced social interaction in the adult male offspring [
16] and increased stereotyped behaviors [
17], and that these animals present several cellular and molecular alterations also observed in autistic individuals [
16,
18]. In these studies, experimental mice exposed to VPA were weaned with other VPA-exposed mice.
To evaluate the contribution of the early social environment to the levels of sociability in adulthood, we weaned VPA mice in VPA only cages (VPA-VPA mice), or mixed them in cages containing animals prenatally exposed to VPA (VPA-SAL mice) and saline-treated animals (SAL-VPA mice) in 2:3 or 3:2 ratio. Mice interacted from postnatal day (PD) 21 to PD60 in their homecage. Given that we assume that social interactions are reciprocal after weaning, VPA-SAL animals, i.e., mice prenatally exposed to VPA who lived with control animals since PD21, received social stimulation from control animals, while VPA-VPA mice did not. These differences in social stimulation in the home cage could affect sociability in adulthood.
We next performed a battery of adult behavioral tests in order to compare VPA-VPA mice with VPA-SAL animals, to reveal if peers could rescue the behavioral phenotype caused by prenatal VPA exposure. In particular, we evaluated ASD-related behaviors and found that early social stimulation can revert the reduction in sociability observed in VPA-VPA mice in the social interaction test, but repetitive behaviors are increased in all VPA-exposed mice (evaluated in self-grooming and spontaneous alternation in the Y maze). The olfactory habituation/dishabituation test was performed to evaluate olfactory function and response to social odors, and VPA-VPA mice showed a specific deficit in social odors investigation. The novel object recognition test was performed to evaluate short-term memory and neophobia, two confounders in the social interaction test, and showed no differences between groups. We then analyzed anxiety- and depression-related behaviors since mood disorders have a high comorbidity with ASD [
19] and can be affected by changes in social stimuli early in life, and found that VPA-exposed animals show less exploratory behaviors and increased depression-related behaviors.
After characterizing the behavioral alterations in our experimental groups, we analyzed brain glucose metabolism in an attempt to identify brain structures involved in these behavioral phenotypes. As our unbiased preclinical PET study pointed to altered function of the piriform cortex, we further analyzed this area and found changes in neuronal activity and dopaminergic function that are comparable to the alterations in sociability observed in VPA-VPA mice.
Discussion
Taken together, our results show that early social stimulation can revert the decrease in sociability observed in VPA-exposed mice, while other behaviors are not affected by this postnatal treatment. In addition, preclinical PET analysis identified specific brain regions whose metabolism is altered by prenatal VPA treatment. Interestingly, most of these regions regain the control levels when VPA animals are reared with SAL mice. Among these brain structures, the piriform cortex (Pir) shows a bilaterally increased metabolism in VPA-VPA animals when compared with SAL-SAL or VPA-SAL mice. We also observed that VPA-VPA mice have more cFos-positive cells in the aPir and increased dopamine turnover in the Pir.
Performing a battery of behavioral tests and measuring a significant number of physiological variables leads to the problem of multiple testing when performing statistical analyses of the results. We reasoned that methods that provide more strict p values (e.g., Bonferroni correction or FDR) could contribute to avoiding false positives, but because of the extensive nature of our analysis, they could also lead to false negatives. VPA treatment, social enrichment, and the use of an outbred strain, all contribute to biological variability, making it difficult to detect meaningful differences. As an alternative, we have employed different strategies in order to minimize the effect of multiple testing on our results. On one hand, we used independent cohorts for juvenile play, PET studies (two cohorts), and HPLC analysis. In addition, adult behavioral testing was performed in three different cohorts, and each behavioral domain was evaluated in more than one test and variable. Moreover, animals employed for c-Fos expression analysis were randomly selected from a behavioral cohort, and a FDR approach was used in PET studies. All these approaches contribute to reducing type 1 error without significantly increasing type 2 error and the number of animals used for experiments.
The development, organization, and function of the social brain circuitry result from the interaction between the child and his/her social environment [
11]. Reciprocal social interactions are required for the integration of social stimuli to engage brain regions involved in reward, motor control, and attention [
39]. In mammals, social interactions take place first with the parents and subsequently with the peers. Altered social environments can worsen genetic or prenatal factors affecting sociability, and we reasoned that, conversely, a rich social environment could counteract prenatal factors affecting sociability. Indeed, most early behavioral interventions employed in treating ASD target social aspects of behavior (e.g., language, social interaction) [
40].
We have previously shown that maternal behavior is not affected by injecting VPA to the dams at GD12.5 [
41], suggesting that maternal care is not contributing to the social deficits observed at weaning (Fig.
2a–c). After weaning, animals only interact with peers in the home cage. We hypothesized that the poor social environment experienced by VPA-exposed mice reared with other VPA-exposed mice could contribute to the reduction in sociability observed in adulthood. In addition, we hypothesized that VPA-exposed animals reared with animals showing more sociability could reverse their phenotype. Our design allowed us to specifically test the role of reciprocal interactions, without the confounding effect of the mother that could emerge from earlier interventions. Moreover, the developmental processes taking place in the rodent brain from PD20-PD25 are similar to those occurring in the 2–5-year-old human brain [
42], when usually ASD symptoms are detected.
We found that post-weaning social enrichment can reverse the altered sociability observed in young VPA-exposed mice, resulting in adult animals showing more social interaction that VPA-exposed animals reared with other VPA mice. Particularly, we show that VPA-VPA animals spend less time sniffing a novel mouse than any of the other groups (SAL-SAL, SAL-VPA, or VPA-SAL). The maze that we have employed in this work, originally described in [
43], consists of small chambers. As a consequence, when animals are located in the lateral chambers they are mostly exploring the cylinder, rarely exploring the maze. Therefore, “sniffing” is the most reliable parameter of sociability since it involves the actual investigation of the social stimulus or the exploration of the control tube. However, it also implies a limitation because we can only provide one measure of social interaction. To evaluate the specificity of this effect, we analyzed other behaviors that are altered by prenatal VPA-exposure. We found that the VPA effects on repetitive behaviors, exploration, and depression-related behavior are not reversed by early social enrichment (Figs.
3 and
4). These results suggest that the previous reported effects of enriched environments on these behaviors [
14,
44,
45] are due to the opportunity to explore a richer physical environment and not to the social stimulation available in such environments.
Alterations in juvenile play are observable at PD21 in mice prenatally exposed to VPA. Our results suggest that there is an additional critical period starting around PD21, when the interactions with peers can set the level of sociability in mice. Although we cannot claim a specific critical period of development for this effect, previous work suggests adolescence (approximately from PD21 to PD42) as a critical period for the establishment of social behaviors [reviewed in [
46]]. Future work could narrow this critical period of postnatal development and identify the mechanisms through which they act. Although VPA and SAL animals show different levels of sociability at PD21 and the only variable we modified was the composition of the cages from PD21 to PD60, we did not specifically evaluate the levels of social stimulation on VPA-exposed animals in VPA-VPA and VPA-SAL groups. Therefore, we cannot rule out the contribution of additional unknown parameters (e.g., microbiota in the fecal matter [
47]) that might affect the development and consolidation of sociability.
Here, we report the effects of VPA and early social stimulation on male offspring. We circumscribed our analysis to males since we did not observe any effects of prenatal VPA on social interactions in female offspring compared with the controls (unpublished results). This is consistent with the reports in CD1 mice [
48], Sprague-Dawley rats [
20], and Wistar rats [
17]. This differential impact of VPA on male and female offspring adds face validity to this ASD model, as this disorder is four times more common in males than in females [
49]. However, the prevalence of autism in children prenatally exposed to VPA is characterized by a 1:1 male to female ratio [
50], and in the VPA model, there is an even effect on both sexes in other ASD-related phenotypes such as anxiety-related behaviors and repetitive behaviors [
17,
48]. Moreover, female VPA offspring exhibit physiological and cellular changes that are consistent with the alterations observed in ASD children [
17,
41,
51]. However, to date, sociability defects in female mice as a result of VPA exposure have not been found. The effects of social enrichment on females prenatally exposed to VPA needs to be further analyzed, as it can contribute to the treatment of girls diagnosed with ASD.
Most of our knowledge on brain circuit abnormalities in ASD individuals comes from functional MRI studies [
5,
52]. However, these reports show numerous inconsistencies on the brain areas involved. We reasoned that our rescued group of animals would be a valuable tool to validate alterations observed in the VPA model. Using [18F]-FDG preclinical PET, we found that brain activity of VPA-SAL animals is more similar to SAL-SAL brains than VPA-VPA brains, as we found alterations in metabolism in the VPA-VPA brain that are not present in rescued animals. These structures appear then as candidate components of the social brain circuitry of the mouse. Interestingly, our results show differences in glucose metabolism in the resting state, similar to studies describing altered connectivity in people with ASD [
5].
The circuitry that regulates social behavior involves many brain structures. In rodents, these structures are mainly components of the olfactory system, such as the anterior olfactory nucleus (AON) and the Pir. Indeed, the Pir controls olfactory perception and is activated by social stimuli [
53], and during the discrimination between familiar and non-familiar social stimuli [
54,
55]. Moreover, the oxytocin receptor-expressing neurons in the Pir mediate odor-driven social learning [
56]. In line with this, we found a bilateral hyperactivation of the Pir in adult VPA-VPA mice, when compared to either control or rescued mice (Fig.
4). Moreover, we found increased cFos immunoreactivity in the layer 2 of the aPir (Fig.
5c), suggesting increased basal neuronal activity in VPA-VPA mice. Although we did not identify the type of neurons expressing c-Fos, the fact that they are localized for the most part in the layer 2 together with their morphology suggest that they are mostly glutamatergic principal neurons.
The Pir projects to the amygdala, basal ganglia, and hippocampus, areas implicated in behavioral output, with the aPir particularly projecting to the anterior part of the amygdala and the mediodorsal thalamus [
57]. Alterations in the aPir function could then result in alterations in the circuitry engaged during social interactions. The Pir shows increased cFos immunoreactivity after exposure to a juvenile conspecific, but it is reduced after a neutral odor [
55]. The increased basal neuronal activity observed in VPA-VPA mice could then preclude the circuitry to respond normally to a social stimulus, leaving the response to non-social odors unaffected (Fig.
2e, f).
In addition, we found increased dopaminergic turnover in the Pir of VPA-VPA mice (Fig.
5i). Cells in the Pir receive input from midbrain dopaminergic neurons [
36], express dopamine receptors [
58], and exhibit dose-dependent responses to dopamine [
59]. Rodent dopaminergic projections and dopamine receptor expression in the brain undergo profound changes during the juvenile (PD21-PD28) and adolescent (PD34-PD49) stages [
60,
61]. Social enrichment could modify a VPA-induced aberrant pattern of maturation of dopaminergic innervations, restoring normal dopaminergic function. Coactivation of dopamine D1 and D2 receptors results in reduced social interaction [
62], suggesting a possible mechanism by which increased dopaminergic function in the Pir could result in diminished sociability in VPA-VPA mice.
In summary, VPA-VPA mice show reduced sociability and increases in glucose metabolism, neuronal activity, and dopaminergic turnover. In turn, VPA-SAL mice show no significant differences in any of these parameters when compared with SAL-SAL mice, displaying a phenotype that was indistinguishable from that of control animals. In addition, VPA-SAL sociability levels are significantly higher and glucose metabolism in the Pir is significantly lower when compared with VPA-VPA mice. These differences are partially explained by the reduction in neuronal activity and dopaminergic turnover observed in VPA-SAL mice, but additional neuronal changes in the Pir may be involved, as it is suggested by the larger differences observed in the PET between both VPA-exposed groups than in the VPA-VPA vs SAL-SAL comparison.
Our preclinical PET results also show alterations in glucose metabolism in other brain structures, the motor cortex (M1/M2), the somatosensory cortex (S2 and S1BF) and the insular cortex. We need to further study these structures to elucidate if they play a role in modulating the behaviors altered in animals prenatally exposed to VPA. It was recently shown that maternal immune activation (MIA) results in cortical patches of reduced cellular activity (cFos-positive cells) in the primary somatosensory cortex (S1) and secondary motor cortex (M2) [
34]. Interestingly, these patches were often present unilaterally, similar to the pattern that we observed (Fig.
4). MIA offspring show behavioral alterations comparable to those observed in VPA-exposed mice, and the specific manipulation of neuronal activity on the S1 can modulate social interaction and repetitive behaviors in MIA offspring [
34]. These results demonstrate a role of the S1 in modulating these behaviors and suggest that the reduction in glucose metabolism observed unilaterally in S1 and M2 in VPA-VPA mice could also contribute to the abnormal behavior of these animals.
Our unbiased analysis of brain activity in a mouse model of ASD adds to previous fMRI studies on brain connectivity and neuronal activity in the acallosal and socially impaired BTBR strain [
63,
64]. BTBR mice showed reduced basal neuronal activity and metabolism in the somatosensory and piriform cortex, when compared with C57BL/6J mice. In addition, brain activity mapping was previously performed in a mouse model of Rett syndrome (the Mecp2 null mice), using a histological technique (Fos labeling) [
33]. Interestingly, that analysis reports reduced neuronal activity in the Pir, somatosensory, and motor cortices. As Mecp2 null mice [
65] and BTBR mice [
66] show deficits in social behavior, our results add to those reports in suggesting that these structures could be central to its regulation.