Cerebellar Oxidative DNA Damage and Altered DNA Methylation in the BTBR T+tf/J Mouse Model of Autism and Similarities with Human Post Mortem Cerebellum

The molecular pathogenesis of autism is complex and involves numerous genomic, epigenomic, proteomic, metabolic, and physiological alterations. Elucidating and understanding the molecular processes underlying the pathogenesis of autism is critical for effective clinical management and prevention of this disorder. The goal of this study is to investigate key molecular alterations postulated to play a role in autism and their role in the pathophysiology of autism. In this study we demonstrate that DNA isolated from the cerebellum of BTBR T+tf/J mice, a relevant mouse model of autism, and from human post-mortem cerebellum of individuals with autism, are both characterized by an increased levels of 8-oxo-7-hydrodeoxyguanosine (8-oxodG), 5-methylcytosine (5mC), and 5-hydroxymethylcytosine (5hmC). The increase in 8-oxodG and 5mC content was associated with a markedly reduced expression of the 8-oxoguanine DNA-glycosylase 1 (Ogg1) and increased expression of de novo DNA methyltransferases 3a and 3b (Dnmt3a and Dnmt3b). Interestingly, a rise in the level of 5hmC occurred without changes in the expression of ten-eleven translocation expression 1 (Tet1) and Tet2 genes, but significantly correlated with the presence of 8-oxodG in DNA. This finding and similar elevation in 8-oxodG in cerebellum of individuals with autism and in the BTBR T+tf/J mouse model warrant future large-scale studies to specifically address the role of OGG1 alterations in pathogenesis of autism.


Introduction
Autism is a clinically complex, heterogeneous, and behaviorally-defined neurodevelopmental disorder characterized by impaired social skills, communication, and repetitive behaviors. Substantial effort has been devoted in recent years to uncover the underlying mechanisms of genomic, epigenomic, proteomic, metabolic, and physiological alterations associated with the disease development. Although the role of genetic factors in autism has been extensively studied, the genetic variations are extremely heterogeneous and their phenotypic penetrance is highly variable in different individuals [1][2][3]. In addition to genetic alterations, mounting evidence indicates a key role of other molecular and physiological abnormalities, including immune dysregulation, neuroinflammation, epigenetics, oxidative stress, and mitochondrial dysfunction [4]; however, the underlying molecular pathogenesis of autism remains elusive.
In recent years, the cerebellum has emerged as one of the key brain regions affected in autism [5]. This is evidenced by several well-established observations indicating the essential role of autism in the development of basic social capabilities [6], its involvement in extensive neural networks that govern the social, communication, repetitive/restrictive behaviors [7], and motor and cognitive deficits impaired in autism [8].
Elucidating and understanding the molecular processes underlying the pathogenesis of autism is critical for effective clinical management and prevention of this disorder. Investigation of these mechanisms using human subjects is desirable; however, epidemiological and imaging studies are not able to address causality or the molecular underpinnings of the disease. Animal models that resemble core human autistic symptoms may substantially overcome the limitations of human studies and have the flexibility to provide important additional clues regarding the etiology and molecular pathogenesis of autism. Several mouse models have been developed to simulate autism symptoms in humans. Among all mouse models, the inbred BTBR T+tf/J mice is one of the most relevant and commonly used animal models to study autism because they exhibit an autism-like behavioral phenotype [9][10][11]. In addition to behavioral similarities, multiple studies have shown that BTBR T+tf/J mice share many similarities in neuroanatomical and physiological features found in individuals with autism [12][13][14].
Escalating evidence suggests that oxidative stress and aberrations in the cellular epigenome, especially aberrant DNA methylation, are key molecular features of autistic phenotype [15,16] that are linked to alterations in glutathione metabolism and folate-dependent trans-methylation and trans-sulfuration pathways [17][18][19][20][21][22]. While alterations causing these pathological signs are of great importance, little is known about the underlying mechanisms responsible for their persistence in the pathophysiology of autism.
Based on these considerations, the goal of this study was to evaluate key molecular alterations postulated to play a role in autism and their role in the pathophysiology of autism using inbred BTBR T+tf/J mice, a strain that exhibits an autism-like behavioral phenotype, including deficits in reciprocal social interactions and social approach, and high levels of repetitive self-grooming behavior [9][10][11] in contrast to C57BL/6J mice, a mouse strain characterized by a high sociability and low grooming behavior [14]. In this study, we demonstrate that DNA isolated from the cerebellum of BTBR T+tf/J mice and from human post-mortem cerebellum of individuals with autism, are both characterized by an increased levels of 8-oxo-7-hydrodeoxyguanosine (8-oxodG), 5-methylcytosine (5mC), and 5-hydroxymethylcytosine (5hmC). The increased 8-oxodG and 5mC content were associated with a markedly reduced expression of the 8-oxoguanine DNA-glycosylase 1 (Ogg1) and increased expression of de novo DNA methyltransferases 3a and 3b (Dnmt3a and Dnmt3b). Interestingly, a rise in the level of 5hmC occurred without changes in the expression of ten-eleven translocation expression 1 (Tet1) and Tet2 genes, but significantly correlated with the presence of 8-oxodG in DNA.

Mouse and Human Cerebellar samples
The frozen cerebellum from male and female BTBR T+tf/J and C57BL/6J mice (8 weeks of age, n55 per gender/strain) were obtained from the Jackson Laboratory (Bar Harbor, ME, USA). The frozen blocks of post-mortem cerebellum from autism individuals (n515) and unaffected control individuals (n515) were obtained from the National Institute of Child Health and Development Brain Tissue Bank for Developmental Disorders at the University of Maryland, and from the Autism Tissue Program at the Harvard Brain Tissue Resource Center, Bellmont, MA, USA. All donors had a confirmed diagnosis of autism based on Diagnostic and Statistical Manual of Mental Disorders and Diagnostic Interview Revised. Autism and control groups were matched, as closely as possible, for postmortem interval, age, gender, race, and cause of death. The demographic data of case-control tissue samples are detailed in James et al. [23]. The analysis of postmortem brain specimens is not defined as human research by the United State Department of Health and Health Services (DHHS) and Food and Drug Administration (FDA) regulations.

Determination of mitochondrial DNA damage
The extent of mitochondrial DNA damage was determined by measuring the mitochondrial DNA and genomic DNA (mtDNA/gDNA) ratio in the mouse DNA samples by using the NovaQUANT Mouse Mitochondrial to Nuclear DNA Ratio Kit (EMD Millipore, Billerica, MA, USA) according to the manufacturer's protocol. Briefly, simultaneous analysis of copy number of the beclin 1 (Becn1) and nebulin (Neb) genes and mitochondrial 12S rRNA gene and mitochondrial DNA fragment trLEU was accomplished by quantitative PCR (qPCR) for each DNA sample, and the mtDNA/nDNA ratio was calculated by using the relative copy number method.

RNA extraction and gene expression analysis using microarray technology
Total RNA was extracted from a cerebellum tissue using RNeasy Mini kits (Qiagen, Valencia, CA, USA) according to the manufacturer's instructions. Gene expression profiles in the cerebellum of BTBR T+tf/J and C57BL/6J mice were determined utilizing Agilent whole genome 8660 K mouse microarrays (Agilent Technologies, Santa Clara, CA, USA). Sample labeling and microarray processing were performed as detailed in the ''One-Color Microarray-Based Gene Expression Analysis'' Version 5.5 (Agilent Technologies) protocol. The hybridized slides were scanned with an Agilent DNA Microarray scanner (Agilent Technologies) at 5 mm resolution. The resulting images were analyzed by determining the Cy3 fluorescence intensity of all gene spots (features) on each array using the Agilent Feature Extraction Software (Version 10.7). The raw data were then uploaded into the ArrayTrack database [24]. The median fluorescence intensity of all the pixels within one feature was taken as the intensity value for that feature. The raw intensity values were then normalized using 75 percentile channel scaling normalization using ArrayTrack. A list of differentially expressed genes was generated with ArrayTrack using a t-test at p-value ,0.05 and a fold change at .2.0.

Functional analysis of significant genes
Ingenuity Pathway Analysis software (IPA, IPA version 9.0; Ingenuity Systems, Redwood City, CA, USA) was used to determine canonical pathways that were enriched for significant mRNA transcripts identified from the t-test analysis. Significance values were calculated based upon a right-tailed Fisher's exact test that determined whether a pathway was overrepresented by calculating whether the genes in a given pathway were enriched within the data set compared to all genes on the array; p,0.05 was selected as the cutoff for significance based on IPA threshold recommendations. Only those pathways with a p-value below the threshold and having more than three representative genes in the data set were considered significant.

Quantitative reverse transcription-PCR (qRT-PCR)
Total RNA (2 mg) was reverse transcribed using random primers and High Capacity cDNA Reverse Transcription kit (Life Technologies, Grand Island, NY, USA) according to the manufacturer's protocol, and cDNA was analyzed in a 96-well plate assay format using the 7900HT Fast Real-Time PCR System (Life Technologies). Each plate contained experimental genes, and a housekeeping gene. All primers were obtained from Applied Biosystems. The cycle threshold (C t ) for each sample was determined from the linear region of the amplification plot. The DC t values for all genes were determined relative to the endogenous control glyceraldehyde 3-phosphate dehydrogenase (Gapdh). The DDC t values were calculated using treated group means relative to strain-matched control group means [25]. The fold change data were calculated from the DDC t values. All qRT-PCR reactions were conducted in triplicate and repeated twice.

Gene-specific DNA methylation
The methylation status of the mouse Ogg1 promoter region was determined by direct bisulfite-sequencing analysis as previously described [27,28].

Statistical analyses
Results are presented as mean ¡ S.D. Data were analyzed by one-way analysis of variance (ANOVA), with pair-wise comparisons being made by the Student-Newman-Keuls method. When necessary, the data were natural log transformed before conducting the analyses to maintain a more equal variance or normal data distribution. Pearson product-moment correlation coefficients were used to determine the relationship between levels of gene expression and 8-oxodG. Pvalues,0.05 were considered significant.

Results
Microarray analysis of gene expression in the cerebellum of BTBR T+tf/J and C57BL/6J mice In recent years a transcriptomic profiling has become a powerful tool for a better understanding of molecular pathways and uncovering the underlying mechanisms in the pathogenesis of various pathological states, including autism. Therefore, gene expression profiles in the cerebellum of BTBR T+tf/J and C57BL/6J mice were examined using high-throughput Agilent whole genome 8660 K mouse microarrays. The microarray gene expression data have been deposited in the NCBI's Gene Expression Omnibus (GEO) database (accession number GSE62594). Unsupervised hierarchical clustering of the gene expression data showed that each mouse strain could be distinguished by its gene expression profile ( Figure 1A). The tight clustering of samples within each mouse strain indicated the high quality data that would allow subtle differences in gene expression to be identified. Principal component analysis utilizing the entire gene expression dataset also showed the relatively tight clustering of each strain and the clear difference between gene expression profiles in the cerebellum of BTBR T+tf/J and C57BL/6J mice ( Figure 1B).
To identify genes that were differentially expressed between BTBR T+tf/J and C57BL/6J mice, a t-test, p,0.05, coupled with a fold-change cut-off.2 was applied. A total of 325 genes, common for male and female mice, were found to be differentially expressed, up-regulated and down-regulated, in the cerebellum of BTBR T+tf/J mice as compared to C57BL/6J mice ( Figure 1C). Despite the fact that differential expression of the majority of these genes may be driven by genomic variations between these mouse strains, a pathway enrichment analysis of the differentially expressed genes showed a strong enrichment in critical cell cycle/ stress-related genes associated with DNA damage, chromatin organization, and cell death signaling, including the XIAP associated factor 1 (Xaf1), WD repeat domain 48 (WD48), forkhead box protein O3 (Foxo3), cyclin-dependent kinase inhibitor 1A (Cdkn1a, p21), cyclin D1 (Ccnd1), poly(ADP-ribose) polymerase 3 (Parp3), and Ogg1 in the cerebellum of BTBR T+tf/J mice ( Figure 1D and Table  S1).

Oxidative DNA damage in BTBR T+tf/J mice and human cerebellum
The Xaf1 gene was the most up-regulated gene in the cerebellum of male and female BTBR T+tf/J mice, with expression levels 238-and 211-times greater, respectively, than in C57BL/6J mice (Table S1). Significant increases were also observed in the p53 apoptotic and redox signaling network. Recently, Park et al [29] reported an association between up-regulation of XAF1 and generation of reactive oxygen species, one of the key pathophysiological events in the molecular pathogenesis of autism [15]. Based on these findings, the level of 8-oxodG in genomic DNA, one of the prevalent and most studied form of oxidative DNA damage caused by reactive oxygen species [30], was assessed in the cerebellum of BTBR T+tf/J and C57BL/6J mice. Figure 2A shows that the level of 8-oxodG in genomic DNA in BTBR T+tf/J cerebellum was slightly, but significantly increased. We also evaluated the level of 8-oxodG in genomic DNA in postmortem cerebellum of individuals with autism, which was substantially greater than in unaffected control individuals ( Figure 2B). In addition to the damage of genomic DNA, the compromise of mitochondrial functioning and integrity of mitochondrial DNA has been reported in individuals with autism [19,31]. Because  mitochondria are a major source as well as a target of free radical damage, the extent of mitochondrial DNA damage was evaluated by determining the mtDNA/ nDNA ratio in the cerebellum of BTBR T+tf/J and C57BL/6J mice. Figure 2C shows that mtDNA/nDNA ratio in the cerebellum of male BTBR T+tf/J mice was significantly lower (p,0.05) as compared to male C57BL/6J mice, while it was not different in female BTBR T+tf/J and C57BL/6J mice.

Expression of base excision DNA repair genes in BTBR T+tf/J mice
In order to investigate the potential mechanism for the persistence of 8-oxodG in genomic DNA in BTBR T+tf/J mice, the expression of key genes involved in the repair of oxidative DNA lesions was examined by qRT-PCR. The expression of base excision DNA repair genes has been shown to be a reliable biomarker of oxidative DNA damage [32]. Figure 2D shows a marked gender-independent decrease in the expression of Ogg1 in the cerebellum of BTBR T+tf/J mice, while other three key genes in the base excision DNA repair pathway, uracil-DNA glycosylase (Ung), apurinic/apyrimidinic endonuclease 1 (Apex1), and DNA polymerase b (Polb) did not differ from C57BL/6J mice. The expression of Ogg1 in the cerebellum of male and female BTBR T+tf/J mice was 70% and 73% lower, respectively, than in age-matched C57BL/6J mice. More importantly, a decrease in the expression of Ogg1 was significantly inversely correlated with the level of 8-oxodG in the cerebellum of BTBR T+tf/J mice (p,0.01) ( Figure 2E).

Status of DNA methylation in BTBR T+tf/J mice and human cerebellum and histone lysine modifications in BTBR T+tf/J mice
It has been reported that one of the consequences of oxidative DNA damage is aberrant methylation of DNA [33,34]. Therefore, the status of global DNA methylation in the cerebellum of BTBR T+tf/J and C57BL/6J mice was evaluated. Because traditional methods do not distinguish between 5mC and 5hmC, we evaluated both forms of methylated cytosines independently. Figure 3A shows that the levels of 5mC and 5hmC in genomic DNA were significantly increased in the cerebellum of both male and female BTBR T+tf/J mice. Importantly, the level of 5mC and 5hmC elevation in genomic DNA in the postmortem human cerebellum of individuals with autism was similarly increased as compared to matched unaffected control individuals ( Figure 3B) with a magnitude substantially greater than in BTBR T+tf/J mice. Importantly, the level of 5hmC in both mouse and human cerebellar genomic DNA was significantly correlated with the extent of 8-oxodG accumulation in DNA ( Figure 3C). In addition to alterations in DNA methylation, the disruption of a normal pattern of covalent histone modifications is another epigenetic change frequently found in human neurodevelopmental diseases, including autism [35]. Hence, the levels of H3K4me3, H3K9me3, H3K27me3, H4K20me3, H3K9ac, and H3K56ac in the cerebellum of BTBR T+tf/J and C57BL/6J mice were evaluated (Figure 4). It is well established that these histone lysine modifications play a crucial role in the maintenance of genomic stability, chromatin assembly and organization, DNA damage and repair, and regulation of gene transcription [36,37]. Figure 5 shows that in contrast to DNA methylation changes, the levels of selective histone lysine methylation and acetylation marks in the cerebellum of BTBR T+tf/J were not different from that in C57BL/6J mice.

Expression of chromatin-modifying and one-carbon metabolism genes in BTBR T+tf/J mice
The status of DNA methylation and histone modifications relies on several factors, especially on the expression and functioning of chromatin modifying genes and proteins, availability of methyl-group donors and the functioning of one-carbon metabolism, and integrity of DNA [35]. Figure 5A shows that the expression of the de novo DNA methyltransferases, Dnmt3a and Dnmt3b, was, respectively, 1.5and 1.8-times greater in the cerebellum of in male BTBR T+tf/J mice than in C57BL/6J mice. The expression of Dnmt1, MeCP2, Tet1, and Tet2 ( Figure 4A) and histone modifying genes ( Figure S1) in the cerebellum of BTBR T+tf/J mice did not differ from those in C57BL/6J mice. . Western blot analysis of histone H3K4, H3K9, H3K27, and H4K20 trimethylation; H3K9, H4K20 dimethylation and H3K9, H3K27, and H4K16 acetylation in the cerebellum of BTBR T+tf/J and C57BL/6J mice. Densitometric analysis of the immunostaining results is shown as percent change in histone modification level in the cerebellum BTBR T+tf/J mice relative to the corresponding values in C57BL/6J mice, which was assigned a value of 100% (mean ¡ SD, n55). The expression of one-carbon metabolism genes did not differ between mouse strains except for the expression of an essential trans-sulfuration pathway gene, cystathionine c-lyase (Cth), in which inhibition is associated with a greater vulnerability to oxidative injury [38] ( Figure 5B). The expression of Cth was reduced by 44% in male BTBR T+tf/J mice as compared to C57BL/6J mice.

Mechanism of Ogg1 down-regulation in BTBR T+tf/J mice
The expression of Ogg1 may be compromised by genetic factors, e.g., single nucleotide polymorphisms, epigenetic factors, e.g., promoter methylation, and an altered function of its transcription regulators, e.g., TSC2. Figure S2-A shows that the Ogg1 gene in BTBR T+tf/J mice is characterized by the presence of four amino acid coding single nucleotide polymorphisms, three in exon 2 and one in exon 3, as compared to C57BL/6J mice. In contrast, a bisulfite sequencing analysis of the Ogg1 promoter region shows no difference in cytosine methylation between BTBR T+tf/J and C57BL/6J mice ( Figure S2-B). Likewise, western blot analysis revealed no difference in the levels of TSC2 and AKT proteins in the cerebellum in these two mouse strains ( Figure S2-C).

Discussion
Autism is a complex disorder that is thought to be the consequence of multiple interdependent events during development [3,4,6]. The transcriptomic analysis of the cerebellum of BTBR T+tf/J mice demonstrates a profoundly dysregulated expression of cell cycle/stress-related genes, mainly p53 apoptotic signaling, DNA damage and repair, and chromatin modifying genes. These findings are in good agreement with previous reports that have convincingly established a dysregulation of these molecular pathways in autistic brain [4,15,16,39]. Oxidative stressinduced damage to DNA and aberrant DNA methylation are considered as key events, since both of them, in addition to genetic factors, may contribute to the heritability of autism. In this study, we demonstrate that DNA isolated from the cerebellum of BTBR T+tf/J mice and post-mortem cerebellum of individuals with autism, is characterized by an increased levels of 8-oxodeoxyguanosine, 5methylcytosine, and 5-hydroxymethylcytosine.
8-oxodG is one of the prevalent and most studied oxidative DNA lesion and oxidative stress marker [30]. Several comprehensive studies have established the important role of the altered cellular redox status in the pathophysiology of autism [15]. The majority of those studies have largely focused on the potential causes of the oxidative stress induction, mainly on the glutathione redox imbalance and mitochondrial dysfunction [17][18][19][20][21][22]. In contrast, the underlying mechanisms associated with the persistence of oxidative lesions in DNA and proteins in the pathogenesis of autism have never been addressed completely beyond the establishing their presence in autistic brain [40]. The results of the present study demonstrate that an increase in the level of 8-oxodG in the cerebellum of BTBR T+tf/J mice may be largely related to the profound inhibition of the Ogg1 expression. In addition to Ogg1 down-regulation, an observed upregulation of Cdkn1a and Ccnd1 genes may contribute to an elevation of 8-oxodG in DNA in BTBR T+tf/J mice [41,42].
OGG1 is a key enzyme preventing the accumulation of 8-oxodG in DNA through recognizing and removing 8-oxodG from DNA and initiating the highly conserved base excision repair pathway [43]. Since OGG1 is the first enzyme in the base excision DNA repair pathway, the accurate DNA repair greatly depends on the ability of OGG1 to remove 8-oxodG. Ogg1 is highly expressed in the brain and has been shown to protect neurons against oxidative DNA damage during development and various pathologic conditions [44,45]. A lack of Ogg1 in the brain resulted in multiple cellular and molecular events, including increased apoptosis and aberrant neuronal connectivity [45], key pathomorphological features of autism. Additionally, it is well-established that the accumulation of 8-oxodG in the genome caused by inhibition of OGG1 is a key event in the pathogenesis of several human pathologies, including cancer [46], neurodegeneration [47], Parkinson's disease [48], and obesity and metabolic dysfunction [49].
Several mechanisms may contribute to the inhibition of Ogg1 expression in BTBR T+tf/J mice. Since the promoter region of the mouse Ogg1 gene contains a strong CpG island and because of an elevated genomic content of 5mC, a number of epigenetic mechanisms may be involved in Ogg1 gene silencing. However, the bisulfite sequencing analysis of the Ogg1 promoter region did not show differences in CpG methylation between the two mouse strains. It is possible that genespecific histone modifications at the Ogg1 promoter region could cause an inhibition of the Ogg1 expression in the cerebellum of BTBR T+tf/J mice; however, despite the fact that we did not observe alterations in the level of global histone modifications. There are other interconnected molecular mechanisms that could contribute to Ogg1 down-regulation. Specifically, it is well-established that TSC2 is a key regulator of the Ogg1 gene [50]. Down-regulation of TSC2 or genetic deficiency of Tsc2 has been reported to cause a marked decrease of Ogg1 mRNA that was accompanied by the accumulation of 8-oxoG in DNA [50]. The loss of TSC2 has been associated with various neuropsychological disorders [51]. Importantly Reith et al. [52,53] and Tsai et al. [54] have reported that genetic Tsc2 or Tsc1 deficiency causes Purkinje cell degeneration and the development of autism-like phenotype. The results of the present study demonstrating only moderate changes in the level of TSC2 protein in the cerebellum of BTBR T+tf/J mice indicate that this mechanism may not be a main cause of the Ogg1 inhibition. Finally, inhibition of Ogg1 may be caused by genetic variations. A computational analysis of the Ogg1 gene, using the GeneNetwork database (www. genenetwork.org), revealed substantial differences in single nucleotide polymorphisms located in coding and non-coding regions of Ogg1 between BTBR T+tf/J and C57BL/6J mice that may explain a reduced gene expression in BTBR T+tf/J mice.
Another important finding of the present study is a substantial increase in the levels of 5mC and 5hmC in the cerebellum of BTBR T+tf/J mice that is also present in the cerebellum of individuals with autism [16,55]. The increased level of 5mC in the cerebellum of BTBR T+tf/J mice found in this study may be attributed to (i) the presence of oxidative DNA lesions [34] and (ii) a marked upregulation of de novo DNA methyltransferases Dnmt3a and Dnmt3b ( Figure 5). Many current reports link the increase in 5hmC content in DNA to the demethylating function of TET enzymatic oxidation of 5mC [56]. The results of the present study demonstrate that an increased level of 5hmC in the cerebellum of BTBR T+tf/J mice occurred without changes in the Tet1 and Tet2 expression. This corresponds to similar findings in the mouse hippocampus during aging [57]. A parallel elevation of 5mC and 5hmC in DNA in the cerebellum of BTBR T+tf/J mice and individuals with autism suggest that the increase in 5hmC level is not due to its role as an intermediate during demethylation of DNA. A significant positive correlation between the 8-oxoG and 5hmC in both mouse and human cerebellum suggests that the mechanism of DNA methylation alterations found in this study may be a consequence of an altered cellular redox status and oxidative stress.
In conclusion, the results of our study demonstrate that oxidative DNA lesions and an altered pattern of DNA methylation are important molecular features of the autism cerebellar phenotype. The data presented herein point that diminished expression of Ogg1 in the cerebellum of BTBR T+tf/J mice caused by single nucleotide variation in the Ogg1 gene might be a driving force that promotes the accumulation of 8-oxodG in DNA. Similar alterations in 8-oxodG in cerebellar tissue from humans with autism and BTBR T+tf/J mouse model warrant future large-scale studies to specifically address the role genetic alterations OGG1 in pathogenesis of autism in human population. Figure S1. The expression of histone-modifying genes in the cerebellum of BTBR T+tf/J and C57BL/6J mice. The gene expression was determined by qRT-PCR as detailed in ''Materials and Methods''. The results are presented as an average fold change in the expression of each gene in the cerebellum of BTBR T+tf/J mice relatively to that in C57BL/6J mice, which was assigned a value 1 (mean ¡ SD, n55). doi:10.1371/journal.pone.0113712.s001 (TIF)