Background

Autism spectrum disorder (ASD) defines a complex group of neurodevelopmental disabilities characterized by a wide range of impairments in social and communicative skills, stereotyped behaviors, and restricted mental flexibility [1]. The neurodevelopmental and neuroanatomical bases of ASD have been confirmed by a number of clinical, neuroimaging and neuropathological studies [1–3]. The most evident abnormality in ASD consists in an early (perinatal) brain overgrowth followed by an arrest of growth during the first year of age [4]. Neuropathological studies on post-mortem samples from ASD patients also showed a number of cellular and cytoarchitectural abnormalities at the level of the cerebral cortex, cerebellum, amygdala and forebrain limbic structures. Anomalies in the cerebellum are the most reproducible neuropathological alterations in ASD patients [3, 5].

A large series of evidence clearly indicates that neuropathological features and behavioral deficits of ASD have a primarily genetic origin. However, the etiology of ASD remains essentially unknown [6, 7]. Transcriptome analysis has also been used to unravel common pathogenic pathways based on the assumption that distinct rare genetic variants or epigenetic modifications affect common biological pathways dysregulated in ASD [6]. Several studies have analyzed genome-wide expression profiles of ASD patients using lymphoblastoid cell lines and blood samples, supporting upregulation of immune genes and downregulation of neurodevelopmental genes as key players in the pathogenesis of ASD (see [8] for a review). Recently, gene co-expression network analysis of autistic brain areas revealed defects in cortical patterning and an enrichment of differentially expressed genes associated with ASD [9].

The homeobox-containing transcription factor engrailed-2 (En2) is crucially involved in the regionalization, patterning and neuronal differentiation of the midbrain and hindbrain [10–15]. Human studies indicated association of two intronic single-nucleotide polymorphisms (SNPs) in the human engrailed-2 (EN2) gene with ASD [16, 17]. Furthermore, the ASD associated A-C haplotype markedly affected EN2 promoter activity when tested with a luciferase reporter assay in rat, mouse and human cell lines [18]. A recent study of the epigenetic evaluation of EN2 in post-mortem cerebellar samples from autistic patients indicated a persistent upregulation of this homeobox gene induced by epigenetic abnormalities in histone methylation patterns that may contribute to Purkinje cell loss in some individuals with autism [19].

Mice lacking the homeobox domain of En2 (En2hd/hd mice; [20], referred to as En2-/-) have been proposed as a model for ASD, due to their complex neuroanatomical and behavioral phenotype. En2-/- mice display cerebellar hypoplasia, including a reduced number of Purkinje cells, and a defect in the antero-posterior pattern of cerebellar foliation [20–23]. The behavior of En2-/- mice is also reminiscent of some features of ASD individuals. Deficits in social behaviors were detected in En2-/- mice, including decreased play and reduced social interactions; locomotor impairment, as well as defective spatial learning and memory, was also reported in these mice [24–26]. Furthermore, we reported dysfunctions in GABAergic interneurons in adult En2-/- mice and demonstrated engrailed protein expression in specific subpopulations of adult hippocampal and cortical interneurons [27].

To unravel recurrent ASD-related neuropathological mechanisms, we took advantage of the En2-/- mouse model and performed genome-wide expression profiling on cerebellar and hippocampal adult tissues. Our transcriptome analysis of the cerebellum and hippocampus of En2-/- mice suggests convergent pathological pathways with ASD, including abnormal synaptic transmission and increased immune response. Furthermore, we provide evidence for a significant enrichment of differentially expressed genes associated to ASD in this mouse model of the disease.

Methods

Animals

Experiments were conducted in conformity with the European Communities Council Directive of 24 November 1986 (86/609/EEC) and were approved by the Italian Ministry of Health and Ethics Committee of the University of Trento. Animals were housed in a 12 hr light/dark cycle with food and water available ad libitum. All surgery was performed under chloral hydrate anesthesia, and all efforts were made to minimize suffering. The generation of En2-/- mice was previously described [20]. The original En2 mutants (mixed 129Sv x C57BL/6 and outbred genetic background) were crossed at least five times into a C57BL/6 background. Heterozygous matings (En2+/- x En2+/-) were used to generate the En2+/+ (wild type, WT) and En2-/- littermates used in this study. PCR genotyping was performed according to the protocol available at the Jackson Laboratory website (http://www.jax.org; mouse strain En2tm1Alj). WT and En2-/- age-matched adult (3 to 5 months old; weight = 25 to 35 g) littermates mice of both sexes were used.

Microarrays

RNAs from dissected hippocampi and cerebella from three adult mice for each genotype were purified using standard column purification according to the manufacturer’s protocol (RNAeasy Mini Kit, Qiagen, USA). RNA quality was analyzed by microfluidic gel electrophoresis on RNA 6000 NanoChips using the Agilent 2100 Bioanalyzer. Only RNA with a high (>9) RNA integrity number was selected and used for subsequent retro-transcription, labeling and array hybridization according to Agilent protocols. Mouse gene expression arrays (Agilent 4X44K slides) were hybridized and scanned with the Agilent microarray station.

Bioinformatics

Intensity values were processed with Agi4x44PreProcess (http://bioconductor.org/packages/2.12/bioc/html/Agi4x44PreProcess.html) using default parameters to remove low-quality probes. Signals were then normalized by means of the quantile normalization method. To evaluate differential expression, we used RankProd (http://www.bioconductor.org/packages/2.11/bioc/html/RankProd.html) [28]. RankProd utilizes the Rank Product (RP) non-parametric method [29] to identify up- or downregulated genes. The RP is equivalent to calculating the geometric mean rank with a statistical method (average rank) that is slightly more sensitive to outlier data and puts a higher premium on consistency between the ranks in various lists. To assess for functional categories overrepresented in the differentially expressed genes, we used DAVID (http://david.abcc.ncifcrf.gov) and Ingenuity Pathway Analysis (Ingenuity Systems, Inc., USA). To focus the functional analysis on brain expressed genes we used, as background for our functional analyses, a list of tissue specific ‘expressed genes’ for both the cerebellum and the hippocampus. Our ‘expressed genes’ lists were obtained by filtering the genes by the normalized expression values and excluding the ones with the lowest expression levels (<10th percentile), and include 13,652 genes for the cerebellum and 13,141 for the hippocampus. The hypergeometric test and the Student’s t-test were computed with R (http://www.r-project.org).

Quantitative PCR

Total RNAs were extracted by Trizol™ reagent (Invitrogen Life Technologies, USA) from dissected hippocampi and cerebella from four WT and four En2-/- adult mice. RNAs were DNAse-treated and purified with RNeasy Mini Kit (Qiagen, USA). cDNA was synthesized from pooled RNAs (2 μg) using the SuperScript™ VILO™ (Invitrogen Life Technologies, USA) according to the manufacturer’s instructions. Individual PCR reactions were conducted in a volume of 20 μl using the KAPA FAST SYBR qPCR kit (KAPABiosystems, USA) according to manufacturer’s instructions. Mouse mitochondrial ribosomal protein L41 (Mrpl41) was used as a standard for quantification as previously shown [30]. Primers (MWG, Germany) were designed on different exons to avoid amplification of genomic DNA. A list of primer sequences is reported in Additional file 1. Each PCR cycle consisted of denaturation for 10 s at 94°C, annealing for 20 s at 60°C and extension for 30 s at 72°C. The fluorescence intensity of SYBR green I was read and acquired at 72°C after completion of the extension step of each cycle. PCR conditions for individual primer sets were optimized by varying template cDNA and primer concentration in order to obtain a single PCR product and amplification efficiency >90%. Relative expression values were calculated using the Pfaffl method [31].

Results

Differential gene expression in cerebellum and hippocampus of En2-/- mice

The cerebellum of En2-/- mice shows Purkinje cell loss and structural abnormalities resembling the neuropathological features observed in ASD patients [20, 21, 23]. To identify genes and pathways altered in En2-/- mice, we initially performed transcriptome profiling of the whole cerebellar tissue. En2-/- and WT cerebella from adult mice were assessed for differential gene expression by microarray and bioinformatical analysis (see Methods). We found 842 differentially expressed genes in the cerebellum of En2-/- mice compared to their control littermates. Among these, 407 and 435 were up- and downregulated, respectively. Alterations in limbic structures have also been shown in the temporal lobes of autistic patients. The main abnormalities were shown in the superior temporal sulcus and the ventral-basal temporal region, both of which are involved in decoding social stimuli and therefore are associated with the social deficits [32–34]. Most importantly, we previously showed anatomical defects in the En2-/- hippocampus that might contribute to the behavioral deficits displayed by these mutants [27]. We therefore combined the cerebellar gene expression profile to that of the hippocampus. We found 862 differentially expressed genes in the hippocampus, among those 378 were upregulated and 484 were downregulated in En2-/- mice compared to their littermate controls. Additional file 2 shows the entire list of genes differentially expressed in the cerebellum and hippocampus of En2-/- mice with the differential expression P value and the percentage of false prediction (pfp) value calculated by RankProd. Given the profound structural and cytoarchitectural phenotype of the En2-/- cerebellum and the variability of the phenotype among individuals, we chose to be more permissive and include genes with smaller differential expression fold changes applying a moderate cutoff to the uncorrected P value [see Additional file 2]. Differentially expressed genes, which are common in the hippocampus and the cerebellum, are summarized in Figure 1 and listed in Additional file 3. Remarkably, En2 was not among the differentially expressed genes. The microarray probe for En2 (A_51_P397876) was designed on the 3′ untranslated region of the gene that is not deleted in the En2-/- locus [20]. Our data confirm the previous reports indicating the presence of a residual transcript from the En2-/- locus [20]. Furthermore, real-time quantitative PCR (qPCR) analysis of En2 expression using homeobox specific primers revealed no expression of the full-length gene in En2-/- mice [see Additional file 4].

Figure 1
figure 1

Venn diagram of differentially expressed genes in En2-/-adult cerebellum and hippocampus. A total of 842 and 862 differentially expressed genes were identified in the adult cerebellum (CB, blue) and hippocampus (Hippo, red) of En2-/- mice, respectively. Among these, 238 show differential expression in both tissues.

Full size image

Functional analysis

To explore the biological processes most relevant to the phenotype of the En2-/- mice, the differentially expressed genes were analyzed through integrated gene-network analysis using the curated Ingenuity Pathway Analysis (IPA) database and the Database for Annotation, Visualization and Integrated Discovery (DAVID). Additional file 5 shows the most significantly enriched disease and cellular function categories obtained with IPA and gene ontology analysis. Functional categories included increased seizure and decreased neurotransmission release in the hippocampus, and decreased cancer-related diseases and development of lymphocytes as cellular function in the cerebellum. To identify statistically significant over-representation of key neurobiological processes, functional annotation analysis was performed with DAVID using a tissue specific list of ‘expressed genes’ as background (see Methods). To verify the tissue-expression pattern of the samples, we first classified the differentially expressed genes based on their tissue expression (P <0.05, calculated using Benjamini multiple testing correction). We only found significant tissue expression terms for En2-/- hippocampus showing significant enrichment in expression of genes related to ‘brain cortex’, ‘brain’, ‘hypothalamus’, ‘eye’ and ‘hippocampus’ [see Additional file 6]. We then analyzed the functional annotation using gene sets from the gene ontology (GO) public databases and our ‘expressed genes’ list as background. Figure 2 shows all GO terms enriched in the differentially expressed genes of En2-/- cerebellum with the gene counts and the relative P value (P <0.05, calculated using Benjamini multiple testing correction). Among the most represented functional categories were several terms related to the major histocompatibility complex (MHC)-mediated immunity and immune response. The GO terms related to the En2-/- hippocampus are shown in Figure 3 with the gene counts and the relative P value (P <0.05, calculated using Benjamini multiple testing correction). Of the enriched terms, many related to the cellular components synapse, synaptic vesicle and neuronal projection. Within the biological processes, the most represented terms were ‘neuronal activities’ and ‘calcium mediated signal’. Interestingly, the keyword ‘phosphoprotein’ was significantly enriched, suggesting a generic alteration in the protein phosphorylation state in the En2-/- hippocampus.

Figure 2
figure 2

Overrepresented gene ontology categories for differentially expressed genes in the En2-/-adult cerebellum. Differentially expressed genes (n = 842) were analyzed for enrichment in gene ontology categories using DAVID with a Benjamini corrected P value cutoff of 0.05. Categories are arranged from most significant and downwards (gray line), for each category the number of genes is indicated by the length of the horizontal bars (gene counts). To highlight distinct ontological categories, bars are color-coded as indicated in the inset to the figure. Data details are also reported in Additional file 5.

Full size image
Figure 3
figure 3

Overrepresented gene ontology categories for differentially expressed genes in the En2-/-adult hippocampus. Differentially expressed genes (n = 862) were analyzed for enrichment in gene ontology categories using public databases with a Benjamini corrected P value cutoff of 0.05. Further details regarding the figure and inset are described in the legend to Figure 2. Data details are also reported in Additional file 5.

Full size image

To assess the functional consequences of En2 ablation, we analyzed enrichment of the differentially expressed genes based on the mouse phenotype from the Mammalian Phenotype Ontology (MPO) project [35], using ToppGene [36]. In the En2-/- cerebellum the only significantly enriched mouse phenotype was ‘loss of dopaminergic neurons’, a term associated with the cerebellar phenotype by the role of En2 in midbrain/hindbrain development and dopaminergic neuron survival [37, 38]. For the hippocampus we found significant enrichment for terms related to seizure and altered synaptic transmission (Figure 4). These data are in accordance with our previously described increase in seizure susceptibility and decrease in GABAergic neuron subpopulations in En2-/- mice [27, 30].

Figure 4
figure 4

Mouse phenotype categories associated with differentially expressed genes in the En2-/-hippocampus. Differentially expressed genes in En2-/- hippocampus were analyzed for enrichment in mouse phenotypes using ToppGene with a corrected P value cutoff of 0.05. Categories are arranged from most significant and downwards (gray line); number of genes in the category is indicated by the length of the horizontal bars (gene counts).

Full size image

Remarkably, when analyzed specifically for ASD-associated genes, the differentially expressed genes showed significant over-representation of known ASD susceptibility genes when compared directly to the repository of the Simons Foundation and Autism Research Initiative (SFARI) database (gene.sfari.org). Since the SFARI database comprises genes related to neurodevelopmental disorders, therefore expressed in the brain, we filtered the SFARI database gene list with our list of ‘expressed genes’ in the cerebellum and in the hippocampus (see Methods for details). We used these SFARI gene lists to calculate enrichment with our differentially expressed genes. Table 1 shows the list of the ASD-associated differentially expressed genes for the cerebellum and the hippocampus. To compute significant enrichment between our differentially expressed genes and the SFARI genes, we employed the hypergeometric test. The statistical analysis indicated significant enrichment only for the hippocampus (P <0.05), whereas no significant enrichment was observed for the cerebellum.

Table 1 Enrichment of autism spectrum disorder (ASD)-related genes in En2 -/- cerebellum and hippocampus differentially expressed genes
Full size table

To compare our findings with the three major genome-wide expression studies on ASD brain tissue [9, 39, 40], we matched the publicly available differentially expressed genes with the same ASD-related gene lists that we used for our analysis. To increase accuracy, we computed, for each study, the hypergeometric test and obtained an enrichment P value that we used for direct comparison. Table 2 summarizes the results of the enrichment analyses, separated in tissue-specific groups, and the comparison between the studies. For Voineagu et al.[9] we re-analyzed the cerebellum data using the GEO2R tool with default parameters and used these results to evaluate the correspondence with our study. The results for cerebellum show a significant enrichment with ASD-associated genes only for the Voineagu et al. study (4.75% enrichment, P = 0.0343). Our study, however, was the only one to display significant enrichment of ASD-related genes (4.24% enrichment, p = 0.0265) in the limbic regions.

Table 2 Correlation of Simons Foundation and Autism Research Initiative (SFARI) database genes with published transcriptome studies in Autism Spectrum Disorder (ASD) brain and our study
Full size table

To validate microarray findings with qPCR analysis, we selected eight and eighteen representative genes from the cerebellum and the hippocampus differentially expressed gene lists, respectively. The selected genes reported differential expression values in microarray experiments ranging between 0.41 fold decrease to 2.5 fold increase. With qPCR, five of the eight selected genes showed statistically significant differential expression in the En2-/- cerebellum [glutamate receptor, metabotropic 5 (Grm5) P = 0.031; neurexin III (Nrxn3) P = 0.032; neuropilin 2 (Nrp2) P = 0.009; v-erb-a erythroblastic leukemia viral oncogene homolog 4 (Erbb4) P = 0.003; calcium channel, voltage-dependent, T type, alpha 1G subunit (Cacna1g) P = 0.003] (Figure 5a). Fifteen genes of the eighteen selected were instead significantly changed in the En2-/- hippocampus [netrin G1 (Ntng1) P = 0.008; Parkinson disease (autosomal recessive, juvenile) 2, parkin, (Park2) P = 0.004; gamma-aminobutyric acid (GABA) A receptor, subunit alpha 4 (Gabra4) P = 0.007; early growth response 2 (Egr2) P = 0.007; sodium channel, voltage-gated, type I, alpha (Scn1a) P = 0.001; neurotrophic tyrosine kinase, receptor, type 3 (Ntrk3) P = 0.0003; phospholipase C, beta 1 (Plcb1) P = 0.0001; cortactin (Cttn) P = 0.0005; synapsin I (Syn1) P = 0.007; fragile X mental retardation, autosomal homolog 2 (Fxr2) P = 0.0004; protein kinase C, beta (Prkcb) P = 0.001; gamma-aminobutyric acid (GABA) B receptor, 1, (Gabbr1) P = 0.03; neuroligin 2 (Nlgn2) P = 0.001; glycogen synthase kinase 3 beta (Gsk3b) P = 0.003; GNAS (guanine nucleotide binding protein, alpha stimulating) complex locus (Gnas) P = 0.01] (Figure 5b). Except for Egr2, in all tested genes the expression differences reported by qPCR correlated with the microarray data (Pearson r = 0.84, P < 0.00005).

Figure 5
figure 5

Quantitative PCR validation of differentially expressed genes. A selected number of differentially expressed genes in the En2-/- cerebellum (a) and hippocampus (b) were validated by qPCR. Relative mRNA expression level (fold expression) as obtained by qPCR performed on whole cerebellum or hippocampus extracts of adult wild type (WT) and En2-/- mice. Correlation of fold expression from qPCR (light color bars) and microarray (dark color bars) results was calculated using Pearson’s Correlation. Except for Egr2, qPCR results for all the evaluated genes showed significant correlation with microarray results (R = 0.84; P <0.05). Values are expressed as each gene/L41 comparative quantitation ratios normalized on the expression of WT (mean ± s.e.m of three replicates from pools of four animals per genotype; P <0.01, Student’s t-test, WT versus En2-/-.

Full size image

Discussion

To date, more than 500 autism-associated genes have been identified (SFARI Gene; gene.sfari.org; updated mar/2013); yet the etiology of ASD remains essentially unknown [6, 7]. The significance of animal models in ASD research has been widely recognized as important for unraveling the molecular, cellular, anatomical, electrophysiological and behavioral consequences of gene dysfunction in ASD. Here, we present a transcriptome analysis in a mouse model of ASD of two brain areas, the cerebellum and the hippocampus, areas that are profoundly affected in ASD patients. Despite the small number of samples used for the microarray analysis and the sample gender heterogeneity, the low genetic variance among individuals allowed a reasonable statistical power for our bioinformatic analysis. Our study revealed that the molecular signature of these two brain regions shares convergent pathological pathways with ASD, including abnormal synaptic transmission and increased immune response. Furthermore, when directly compared to the repository of the SFARI database (gene.sfari.org), our differentially expressed genes in the hippocampus show an enrichment of ASD-associated genes significantly higher than previously reported [41].

Transcriptome analysis has been employed to unravel common pathways based on the assumption that the core phenotypes of ASD may be caused by convergent molecular mechanisms [6]. Several studies have analyzed genome-wide expression profiles of lymphoblastoid cell lines and blood samples from ASD patients, pointing to an upregulation of immune genes as key mechanisms in the pathogenesis of ASD [8]. Despite the limited source of brain tissue samples from ASD cases and the technical restrictions, studies of ASD brain transcriptome are emerging as strategic for uncovering functionally relevant alterations in gene expression. A previous microarray study found alterations of glutamatergic neurotransmission in ASD cerebellum [39], and expression profiles from ASD patient temporal cortices showed upregulation of genes involved in innate immune response and downregulation of several neurodevelopmental genes [40]. Moreover, the transcriptome profiles from three different brain regions (frontal cortex, temporal cortex and cerebellum) of nineteen autism cases and seventeen controls were investigated recently using classical differential expression analysis and a network-based approach [9]. These analyses showed upregulation of genes involved in immune response and downregulation of genes involved in synaptic function and vesicular transport [9]. Our results are in accordance with these findings. Using gene ontology enrichment, integrated gene-network analysis and mouse phenotypes analysis, we report significantly enriched functions and pathways that were previously associated to ASD [42]. In detail, we found increased immune response and major histocompatibility complex-related immunity in the En2-/- cerebellum; decreased and abnormal neurotransmission and increased seizures in the En2-/- hippocampus [see Additional file 5 for details]. Moreover, by direct comparison with the SFARI repository of ASD-related genes, we show that the gene expression changes observed in the En2-/- hippocampus were significantly enriched in ASD-related genes. Furthermore, the proportion of ASD-associated genes enrichment in En2-/- hippocampus was significantly higher than previous studies (Table 2) when compared with Voineagu et al.[9], likely the most comprehensive transcriptome study of ASD post-mortem brain to date. In the case of the cerebellum, in contrast to Voineagu et al.[9] we did not find significant enrichment of ASD-associated genes in En2-/- mice. Such difference could be the result of the complex structural and cytoarchitectural abnormalities in En2-/- cerebellum [20, 21] and the consequent phenotypical variability, or could simply reflect differences between mouse and human phenotypes, as the incidence of cerebellar hypoplasia was not reported in the diagnostic criteria used in the study [9]. Remarkably, EN2 was among the differentially expressed genes found in Voineagu et al.[9], confirming our evidence about the role of En2 in the neuropathology of ASD, and in anterior brain structures [27].

Among the differentially expressed genes, Grm5, Nrxn3 and Scn1a are of particular interest for ASD. Grm5 encodes mGluR5, a G-protein coupled receptor for the neurotransmitter glutamate [43]. In a recent study, mGluR5 has been shown to participate in the pathogenesis of fragile X syndrome (FXS) while genetic downregulation of Grm5 was able to compensate for some of the symptoms in a mouse model of FXS [44]. Furthermore, Grm5 was shown to be downregulated in hippocampal neurons lacking Shank3, another ASD-associated gene [45]. These data support a central role for Grm5 in neurobiological pathways related to ASD pathogenesis. Our results show an increased expression of Grm5 in the cerebellum of En2-/- mice, suggesting a role of Grm5 in the cerebellar phenotype of these mice. The contribution of Grm5 and its interaction with Fmr1 in the En2-/- hippocampus remains to be established and could open new perspective of pharmacological and genetic rescue of the ASD-related phenotype of these mice.

Nrxn3 encodes neuronal adhesion proteins of the Neurexin (NRXN) family. NRXNs are presynaptic cell adhesion proteins that form trans-synaptic complexes with their postsynaptic counterpart neuroligins (NLGNs) and have important roles in synapse development and function [46]. Recently, a report of hemizygous and de novo deletions involving NRXN3 in ASD families provided strong support for a causative link between the loss of NRXN3 and the development of ASD [47]. Our results of an increased expression of Nrxn3 in the cerebellum suggest alterations in Purkinje cell synaptic formation, where NRXNs have been shown to participate to the formation of glutamatergic synapses through interaction with Cerebellin 1 precursor protein (also downregulated in the En2-/- cerebellum) and GluR∂2 [48, 49].

Scn1a encodes the voltage-gated sodium channel alpha subunit. De novo null mutations in SCN1A result in severe myoclonic epilepsy of infancy [50]. SCN1A mutations have been associated to a number of neurological disorders, including generalized epilepsy with febrile seizures plus, Dravet syndrome, borderline myoclonic epilepsy in infancy, intractable childhood epilepsy with generalized tonic-clonic seizures, familial hemiplegic migraine, and a number of cryptogenic focal and generalized epilepsies. Recently, de novo mutations in SCN1A have been associated with ASD [51], and a report of a recognized mutation in SCN1A suggests a wide phenotypic variation of the gene mutations causing a variety of neurologic disorders, including ASD [52]. In mice, heterozygous loss-of-function mutation in Scn1a (Scn1a+/-), reproduces several of the symptoms associated to the human mutation, such as thermally induced and spontaneous seizures, premature death, ataxia and sleep disorder [53, 54]. Scn1a+/- mice show both cognitive deficits and autistic traits that are caused by impaired GABAergic neurotransmission and can be rescued by drug treatment. Scn1a down-regulation in the En2-/- hippocampus could contribute to the abnormal excitability and altered GABAergic neurotransmission shown in these mice by our previous studies [27, 30]. Pharmacological rescue of the hippocampal phenotype in the En2-/- with GABAergic drugs is currently under investigation.

Anomalies in the cerebellum are the most reproducible neuropathological alterations in ASD patients. Several cerebellar abnormalities have been observed in mouse models of both En2 gain- and loss-of-function. Ectopic overexpression of En2 in Purkinje cells during late embryonic and postnatal cerebellar development results in reduced cerebellar volume and loss of Purkinje cells and other cerebellar neurons [55, 56]. Interestingly, En2 knock-out causes defective cerebellar patterning, reduced Purkinje cell number and abnormal dendritic foliation [10, 57], indicating that alterations in En2 expression levels during development cause similar phenotypes. Furthermore, deficits in social behaviors as well as defective spatial learning and memory were also reported in En2-/- mice [24–26]. A recent epigenetic analysis of EN2 promoter methylation in the cerebellum of ASD individuals indicated hypermethylation of the promoter region and persistent upregulation of the gene. The authors report that promoter hypermethylation is normally associated with a decrease in gene expression and suggest the possibility of a developmental mechanism intended to support downregulation of EN2 during perinatal development [19]. Taken together, this evidence suggests that an overall imbalance in EN2 expression may be relevant for ASD pathogenesis, as it could produce alterations in critical brain functions. Comparable evidence of a similar dosage effect has been reported in the case of mutations of other genes critically involved in gene expression regulation and maintenance of synaptic and neuronal homeostasis, such as MECP2 and ARX[58, 59]. It remains to be established whether En2 overexpressing mice display abnormal behaviors relevant to autism. Microarray data have been produced for En2 overexpressing Purkinje cells on a different platform; however, the results overlap only marginally with the herein reported study [60].

Conclusions

Using transcriptome analysis, we identified over 800 genes differentially expressed in the cerebellum and hippocampus of En2-/- mice. Despite the small number of samples used and the relatively small statistical power, our study is the first to analyze molecular changes occurring in two brain structures with neuroanatomical alterations relevant to ASD in a mouse model of this disease. Our bioinformatic analysis of the molecular signature of En2-/- cerebellum and hippocampus shows a significant convergence of neurobiological pathways previously linked to ASD pathology in brain samples from ASD patients. Overall, the present study points to a strong impact of transcriptome analysis on mouse models for identifying neurobiological pathways commonly altered when ASD genes are disrupted in a human patient and in a mouse model alike. Furthermore, together with the frequent association of cerebellar neuroanatomical alterations to the neuropathology of ASD, our molecular analysis suggests a contribution also for the hippocampus, where molecular changes relevant to ASD may occur also in human patients. This notion is supported by the consistent enrichment of ASD-related genes in the En2-/- hippocampus compared to the cerebellum and to other similar studies performed on ASD patient tissue samples [41]. Finally, our results confirm the En2-/- mouse model of ASD as a valuable tool for investigating neuroanatomical, behavioral, as well as molecular alterations related to ASD.

Availability of supporting data

The data sets supporting the results of this article are available in the GEO repository, GSE51612.