Background
Autism spectrum disorders (ASD) represent an etiologically heterogeneous group of disorders that are believed to be caused by a myriad of genetic and environmental factors. ASD patients typically exhibit symptoms of repetitive behavior, impaired social interactions and deficits in social communication. These patients also frequently exhibit intellectual disability, epilepsy, attention deficit hyperactivity disorder (ADHD), sleep disturbances [
1‐
3], and anxiety [
4,
5]. Additionally, some ASD patients may exhibit a reduced ability to be fear conditioned [
6].
It is difficult to trace the pathology of a heterogeneous disorder such as ASD to a single brain region. However, the presence of significant social and emotional abnormalities among the human ASD population indicates atypical amygdala function including the basolateral complex which plays an important role in processing emotional and social cues [
7‐
13]. Collectively, more research is needed to further elucidate the underlying molecular and cellular basis of amygdala dysfunction in ASD.
Twin and family studies have indicated that there is a strong genetic basis for ASD [
14‐
18]. The genetics of ASD is highly complex, involving multiple genes with a high degree of genetic variation [
19,
20]. Notably, a recent study found that siblings with ASD often possess very different genomic mutations, indicating that even within a single family, the cause of ASD can be heterogeneous [
21]. Although the genetic basis of ASD is well documented, the recent increase in clinical cases of idiopathic ASD indicates that environmental risk factors might also have an important contribution, either by causing new mutations or by increasing the risk for ASD in a genetically predisposed individual [
22‐
24].
Studies utilizing animal models of ASD have significantly increased over the past decade, allowing researchers to gain a better understanding of the neurobiology of ASD. One such animal model is the valproic acid (VPA) model of autism [
25‐
27]. This model is based on the discovery that administration of the anticonvulsant drug VPA during the first trimester of pregnancy increases the likelihood of having children with ASD and intellectual disability [
22‐
24,
28,
29]. In this animal model, pregnant dams are administered a single dose of VPA on or around embryonic day 12.5 of gestation, during the time of neural tube closure. The resultant progeny display anatomical and behavioral abnormalities similar to human ASD [
26,
30‐
32], including deficits in social interaction, increased levels of anxiety, and abnormal fear learning [
33‐
36]. Several independent groups have speculated that synaptic deficits may contribute an important role in the causal mechanism of ASD. Synaptic abnormalities have been observed in fragile X [
37,
38], Rett syndrome [
39], Angelman syndrome [
40], and tuberous sclerosis complex [
41‐
43]. Proteins required for normal synaptic functioning have also been implicated in ASD such as Shank, MeCP2, Reelin, and Neuroligin. [
43‐
47]. These proteins contribute an important role to synaptic plasticity and learning and memory.
In the current study, we aimed to understand the molecular basis of amygdala dysfunction, in animals exposed to VPA in utero. We performed genome-wide gene expression profiling on amygdala tissue obtained from adult rats that were exposed to VPA in utero. We identified Homer1a, which is a dominant negative regulator of the critical synaptic scaffolding protein Homer1, to be significantly upregulated in the basal and lateral amygdala (BLA) of VPA-exposed animals. Homer1a is especially interesting since it interacts directly with important synaptic proteins such as metabotropic glutamate receptors (mGluR5) and Shank, which have previously been associated with ASD [
48‐
54]. Additionally, Homer1a contributes a critical role in plasticity and fear learning [
55‐
58] indicating that its dysregulation could underlie fear learning abnormalities observed in VPA-exposed animals. Therefore, we hypothesized that overexpression of Homer1a in the BLA could impair amygdala-dependent phenomena such as fear learning or social interaction behavior [
11,
59‐
63]. To test our hypothesis, we used a viral-mediated approach and overexpressed Homer1a in the BLA of naive animals. We found that overexpression of Homer1a in the BLA impaired fear learning and reduced social interaction in the animals but had no influence on locomotor behavior or anxiety as measured by the open-field test. Our findings are intriguing in part because the Homer1 gene has been previously associated with the human ASD population [
64] thus underscoring that an environmentally induced animal model of ASD could provide novel avenues for elucidating the neurobiological mechanisms of ASD.
Discussion
A growing body of literature has identified a range of autism-like behavioral and synaptic abnormalities in animals exposed to VPA in utero. This includes a previous study from our lab where we identified fear learning and social abnormalities in VPA-exposed animals. In this study, we identified Homer1a as one of the molecules that was significantly upregulated in the BLA of VPA-exposed animals at both the mRNA and protein levels. We hypothesized that upregulation of Homer1a in the BLA might be contributing to the behavioral abnormalities observed in VPA-exposed animals. To test our hypothesis, we used a viral mediated approach to overexpress Homer1a in the BLA of naïve animals. Our results revealed for the first time that overexpression of Homer1a within BLA neurons was capable of impairing auditory fear conditioning and social interaction. These results are very interesting in part because we were able to demonstrate a possible causal mechanism underlying fear conditioning and social interaction abnormalities observed in an animal model of autism.
Homer is a family of scaffolding proteins found at the postsynaptic density. Homer1b/c is a longer isoform of the family which interacts with a number of proteins at the postsynaptic density. It has an EVH1 domain and a proline-rich containing motif at the N-terminus, which binds to various scaffolding and signal transduction molecules, such as type I mGluR, IP3 receptors (IP3R), Shank, transient receptor potential canonical (TRPC) family channels, and dynamin3 [
49,
78‐
81]. The C-terminus of long isoforms of Homer1 contain a coiled-coil domain followed by two leucine zipper motifs which promotes homomeric or heteromeric interactions with other long isoforms of Homer to form a network and stabilize the integrity of the postsynaptic density [
50].
Homer1a is an immediate early gene (IEG) that codes for a shorter isoform of Homer1, which lacks the coiled-coil domain and the ability to interact with other Homer1 molecules. This allows Homer1a to act as a dominant negative regulator by disrupting the interaction of the long isoforms and therefore allows Homer1a to reorganize the postsynaptic density. Homer1a is also involved in intracellular calcium homeostasis, receptor trafficking, gene transcription, signal transduction, and homeostatic synaptic downscaling [
48,
57,
82,
83]. Homer1a also plays a role in pain plasticity by protecting against chronic inflammatory pain without affecting the basal pain threshold [
76]. Additionally, Homer1a contributes an important role in learning and memory, and
Homer1a knockout animals exhibit reduced fear memory acquisition as well as reduced short-term memory [
56]. Overexpression of Homer1a inhibits dendritic spine morphogenesis as well as reduces the size of PSD95 clusters,
N-methyl-
d-aspartate (NMDA) receptor clusters and surface levels of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors [
84]. These findings indicate that either a deficit or excess of Homer1a could be pathological and prevent normal synaptic functioning.
There is evidence indicating that dysregulation of Homer1 may be occurring in a sub-population of ASD patients. For example, a study comprised of a population diagnosed with non-syndromic autism identified a significant number of genes involved in the mGluR5 pathway including Homer1 [
64]. Specifically, single nucleotide variants (SNV) were identified, which localized to the EVH1 domain and 3′ untranslated region (UTR) region of Homer1. A separate postmortem study found reduced Homer1 protein expression in the frontal cortex of ASD patients [
85]. Moreover, different animal models of autism have indicated that Homer1 and Homer1a function may be dysregulated in these models. One study found Homer1a to have an increased interaction to mGluR5 receptors in the hippocampus, which would disrupt Homer1 scaffolds at the synapse of fragile X mice. Increased interaction of Homer1a to mGluR5 in turn enhanced mGluR5-dependent long-term depression (LTD) [
38]. Interestingly, a different study found an increased interaction of mGluR5 to the long isoform of Homer1, and it was found that this increased interaction also enhanced mGluR5-dependent LTD in an Angelman syndrome mouse model [
86]. Surprisingly, in the Angelman Syndrome study, changes in the coupling of mGluR5 receptors to Homer proteins were opposite to those seen in fragile X. Nevertheless, these studies underscore the significant role of Homer1a in the neurobiological mechanisms leading to ASD.
Our microarray data also revealed other genes such as
Mef2C,
gfap,
grxcr1, and
sgk1, which were dysregulated in VPA-exposed animals (complete gene list in Additional file
1.). Homer1a’s previously established role in learning and its association with human ASD made it an attractive candidate to pursue in the present study. However, it will be interesting to investigate the role of these other genes in the future to determine how they may contribute to the aberrant behavioral and synaptic abnormalities seen in the VPA-exposed animals. We propose that dysregulation of many genes, which include
Homer1a, gives rise to the ASD phenotype seen in VPA-exposed animals, and it would be unlikely that Homer1a alone is solely impairing fear conditioning or social interaction. Our experimental data clearly indicate that Homer1a is consistently upregulated in neurons from the BLA of VPA-exposed animals and that when we overexpress Homer1a within the BLA utilizing viral mediated gene delivery, we observe impairments in fear conditioning and social interaction behavior compared to the control groups. However, an important caveat of these viral-mediated gene delivery experiments is that they do not provide the ability to precisely adjust Homer1a levels to mirror what is occurring exactly within VPA-exposed animals.
It is well known that the amygdala can modulate social behavior and is a hub for brain networks that support social life. For example, bilateral lesions of the amygdala in monkeys can reduce social behavior and increase social phobia [
77] and neonatal amygdala lesions suppressed social interactions in adult rats [
87]. A recent study demonstrated a critical role for BLA neurons that project to the hippocampus in bi-directionally modulating social behavior. This study demonstrated that over activation of these BLA projection neurons could reduce social interaction [
88]. However, the molecular mechanisms in the amygdala that influence social behavior still remain largely unknown. For the first time, we demonstrate that overexpression of Homer1a in BLA neurons impairs social interaction in naïve animals.
Valproic acid is a histone deacetylase (HDAC) inhibitor and is prescribed for the treatment of epilepsy [
89‐
91]. The mechanism by which VPA acts at the molecular level is currently not fully understood. Pups exposed to VPA
in utero exhibited increased total brain, brain-derived neurotrophic factor (BDNF) expression and abnormal development of serotonergic neurons in the dorsal raphe nucleus [
92,
93]. Various biochemical studies indicate that VPA can suppress neuronal activity by blocking sodium and calcium channels and enhance the functioning of the inhibitory neurotransmitter, gamma-aminobutyric acid (GABA), in the brain [
94,
95]. Further, in an attempt to understand the mechanism of VPA-induced toxicity, several studies found dysregulated gene expression related to organ morphogenesis and neural tube defects [
96‐
100], and these affects have been attributed to VPA’s ability to inhibit histone deacetylase [
101‐
103]. Since HDAC plays an important role in regulating transcription during fetal development [
45,
104,
105], it is possible that VPA may induce abnormal gene expression during embryogenesis, causing autism-like behavioral impairments. For example, a recent rodent study demonstrated that inhibition of HDAC
in utero is sufficient to cause autism-like phenotypes including sociability deficits in exposed offspring [
106]. Further research needs to be conducted to understand how exposure to VPA
in utero is causing overexpression of Homer1a.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
AB and JAL generated and purified the viruses. AB and JAL performed the viral infusion surgeries. AB performed the behavioral assays including fear conditioning, social interaction, and open field. AB, AH, and AOS helped in sectioning the brain and placement analysis. AB and AH performed the LMD and qRT-PCR, and AB performed western blots. AB and JEP conceived the study and participated in its design and coordination and drafted the manuscript. All authors read, edited, and approved the final manuscript.