Significant strides have been achieved in unraveling the genetic risk architecture of ASD . Thus, it is now estimated that up to 30 % of cases harbor either (i) identifiable chromosomal microdeletions and duplications (copy number variants) that often encompass many hundreds of kilobases of sequence on the linear genome (7–20 % of cases), or (ii) single gene disorders such as Fragile X, Rett Syndrome and others that in toto account for 5–7 % of all cases, and another (iii) 5 % of cases are due to metabolic diseases including mitochrondrial disease, phenylketonuria, adenylosuccinate lyase deficiency etc (Schaaf and Zoghbi 2011). This still leaves 70 % of cases with no straightforward genetic explanation, albeit genetic approaches such as whole exome and whole genome sequencing are beginning to make additional inroads in the complex field of ASD genetics, mostly by identifying additional rare variants of potentially high disease risk (Gratten et al. 2013; Shi et al. 2013) . There are also ongoing refinements in mapping some of the common (mostly single nucleotide) polymorphisms that contribute to the genetic risk of ASD and other psychiatric and neurodevelopmental disorders (Smoller et al. 2013).

In addition to the above described genetic risk structure of ASD—there is both indirect and direct evidence for epigenetic component(s) in the etiology of the disorder. At the population level, twin studies have almost exclusively been used to demonstrate substantial genetic contribution to ASD etiology. However, it is important to underscore that the concordance rates are not exceeding 50 % in monozygotic twin studies in ASD and related disease (Steffenburg et al. 1989; Bailey et al. 1995; Chang et al. 2013)—providing ample evidence for the role of environmental factors, and for epigenetic processes in the etiology of ASD. The framework for the epigenetics of ASD includes an increasing list of deleterious mutations and DNA structural variants directly resulting in dysregulated chromatin, including several portions of chromosomes 15q and 7q that are classical sites for genomic imprinting, as defined by allele-specific DNA methylation signatures differentially regulated on the maternally vs. paternally derived chromosomes (Schanen 2006). Therefore, it is entirely possible that disruption of gene expression via epigenetic mechanisms not reflected in the primary nucleotide sequence, or so called ‘epialleles’, either due to de novo dysregulation in very early development, or by inheritance and carry-over of epigenetic information through the parental germline, plays an important role in some causes of ASD and other psychiatric disorders .

Here, we define epigenetic heritability simply as a phenotype or predisposition that is transmitted from one generation to the next, by some sort of molecular information in the germ cell, that is not encoded by the base pair DNA sequence of the (transmitted) genome. It is often speculated that ‘abnormal’ epigenetic decoration of the haploid germ cell genome (including DNA and histone modifications, see below) contribute to heritable risk of common (including psychiatric) disorders but convincing evidence, or even proof, of ‘true epigenetic heritability’ in disorders such as ASD is extremely difficult to accomplish with present day technologies, because it would, among others, require demonstration that DNA structural variants either in cis (at the site of the epigenetic defect) or trans (other locations in the genome) were not driving the chromatin alterations. Nonetheless, there is an increasing amount of correlative evidence in support of epigenetic heritability in humans, including the role of the ancestral nutritional environment in metabolic disease risk in children and grandchildren (Rando 2012). In the following sections, we summarize some of the most interesting findings as it pertains to the possibility for disease-relevant epigenetic heritability :

The list of mutations carrying a high risk for ASD, even as singular genetic factors, includes an increasing number of genes encoding a chromatin regulatory protein . This is not too surprising, because chromatin remodeling and proper assignment of epigenetic marks is of fundamental importance for brain ontogenesis and a key control point in the stepwise transition from pluripotency to neural precursor to terminally differentiated neurons and glia (Ho and Crabtree 2010), and involved in developmental events such as neuronal migration and connectivity formation (Fuentes et al. 2011). On the other hand, ASD-associated chromatin disorders were initially known only in the context of embryonic defects and multi-organ syndromes, with some of the more recently discovered gene defects affecting selectively the brain without significant involvement of peripheral organs. Well known examples of the former type of gene defects include Rubinstein-Taybi syndrome (RSTS) 1 (Online Mendelian Inheritance of Man, OMIM 180849) and 2 (OMIM 613684) (Roelfsema and Peters 2007), as well as the Immunodeficiency, centromere instability, facial anomalies (ICF1) mental retardation syndrome (OMIM 242860) (de Greef et al. 2011; Ehrlich et al. 2008), among others. The most well known example of the latter type of gene defects includes Rett Syndrome, an X-linked neurological disorder of early childhood (OMIM 312750). To date, deleterious mutations in over 50 different chromatin regulators have been associated with psychiatric disorders including ASD and schizophrenia (Ronan et al. 2013) .

Chromatin remodeling and epigenetic mechanisms are involved in developmental events such as neuronal migration and connectivity formation (Fuentes et al. 2011) . Furthermore, there is increasing evidence that the normal course of maturation and aging is associated with changes in the brain’s epigenome . On the one hand, this is an attractive hypothesis given that there are widespread age-related changes in gene expression in the cerebral cortex , including downregulation of many neuronal genes (Erraji-Benchekroun et al. 2005; Tang et al. 2009). However, in contrast to the accumulation of somatic mutations and other structural brain DNA changes that affect promoter function during aging (which are likely to be irreversible) (Lu et al. 2004), most or perhaps all epigenetic markings studied to date are now thought to be reversible, and there is no a priori reason for unidirectional accumulation of a specific epigenetic mark in aging brain chromatin. Nonetheless, an increasing body of literature indicates that a substantial reorganization of the epigenome occurs during postnatal development and aging. Human cerebral cortex, for example, shows complex and gene-specific changes in levels of methyl-cytosine (mC5; cytosines are methylated at the carbon 5 position), with a steady rise at many promoters that continues into old age in conjunction with subtle changes (mostly a decline) in gene expression (Siegmund et al. 2007; Hernandez et al. 2011).

The increasing list of monogenic forms of neurodevelopmental disease due to mutations in genes encoding specific chromatin remodelers speaks to the importance of these mechanisms for brain ontogenesis and early maturation. It is interesting to note that DNA and histone methylation signatures show, on a genome-wide scale, the most dramatic changes during prenatal development, and the transition phases from perinatal period to early infancy and the perinatal period; in striking contrast, epigenetic changes during subsequent periods of maturation and aging, including puberty and various phases of adulthood, are comparatively minor (Siegmund et al. 2007; Numata et al. 2012; Shulha et al. 2013; Cheung et al. 2010). Cell-type specific epigenome profilings confirmed that these developmental changes are not explained by the various shifts in neuron-to glia ratios and other changes in cell type composition and instead, point to large scale chromatin remodeling in cortical neurons during prenatal and early postnatal development (Cheung et al. 2010; Shulha et al. 2013; Siegmund et al. 2007).

Thus, one could conclude that immature neurons are, both in utero and for the first few months or years after birth, particularly vulnerable to epigenetic pertubations, and some of these could have played a role in the etiology of some cases on the autism spectrum. This hypothesis is further supported by the observation that prenatal exposure to certain types of drugs, including the short chain fatty acid derivative and histone deacetylase inhibitor, sodium valproate, is associated with a 2–3 fold increase in the risk of the offspring to develop ASD (Christensen et al. 2013). Because epigenetic mechanisms literally serve as a molecular bridge linking the genome to the environment, it is worth noting in this context that a recent large twin cohort attributed the estimated risk of shared in utero environment at 30–80 % for ASD, exceeding the estimated genetic risk of 14–67 % (Hallmayer et al. 2011).

ASD, like many other psychiatric disorders, lacks a unifying neuropathology and is probably comprised of a diverse set of syndromes of considerable genetic and etiologic heterogeneity . Interestingly, however, studies even in fairly small postmortem brain cohorts with typically less than 20 cases and equal number of controls have shown significant, disease-associated changes in RNA and protein levels. For example, there is literature reporting downregulated GABA_A and GABA_B receptor expression and ligand binding across multiple areas of the cerebral cortex and cerebellum , in conjunction with altered expression of GABA synthesis enzymes and GABA neuron-specific peptides in cerebral cortex, hippocampus and cerebellum, reviewed in (Blatt and Fatemi 2011). As in other conditions, including schizophrenia (Akbarian and Huang 2006), changes in the GABAergic transcriptome appear to affect a significant portion of cases on the autism spectrum, albeit the reported RNA alterations are not sufficiently consistent to reach the level of significance in all of the postmortem studies (Huang et al. 2010). Nonetheless, this work in the clinical tissue samples provided one of the cornerstones for the popular hypothesis that alterations in the balance of excitatory and inhibitory (E/I) activity play a critical role in the pathophysiology of a substantial number of cases on the autism spectrum, affecting cortical inhibitory (GABAergic) circuitry and synchronization of electrical activity across widespread brain regions in the autistic brain (Uhlhaas and Singer 2012; Rubenstein 2010) .

The complete set of gene expression changes in the autistic brain are likely to go far beyond these GABAergic RNAs. For example, recent microarray studies identified a diverse set of neuronal genes and noncoding RNAs, many of which positioned in genomic loci conferring genetic ASD susceptibility and that showed altered expression in prefrontal and temporal cortex or cerebellar cortices (Voineagu et al. 2011; Ziats and Rennert 2013), with additional gene expression changes indicative of a dysfunction of immune regulation in at least some of the disease cases (Voineagu et al. 2011). Given that transcriptional regulation is closely associated with dynamic changes in chromatin structure and function (Lee and Young 2013), it would not be too surprising that these gene expression alterations in the diseased brain are associated with epigenetic changes in cis-regulatory sequences such as transcription start sites.

However, one challenge for epigenetic studies in brain is the enormous cellular heterogeneity of the postmortem tissue, with its mixed population of glutamatergic and GABAergic neurons, mature oligodendrocytes and astrocytes and their precursors as well as endothelial cells. Furthermore, some brains from subjects on the autism spectrum show evidence for inflammation , including activation and proliferation of microglia as the brain’s immune surveillance cells (Theoharides et al. 2013; Suzuki et al. 2013; Tetreault et al. 2012). This is an important problem given that cell-type specific differences in DNA and histone methylation landscapes are, on a genome-wide scale, far greater than, for example, developmentally regulated changes within a specific cell type, or species-specific regulation in the primate brain (Shulha et al. 2013, 2012b; Cheung et al. 2010) . Thus, conventional chromatin assays, which are designed to detect and quantify DNA methylation and histone modifications requiring an input material between 10³–10⁸ nuclei (Huang et al. 2006; Adli and Bernstein 2011), could be severly confounded by changes in glia-to-neuron ratios that could show considerable fluctuations across different developmental or disease states. To date, many studies exploring epigenetic dysregulation of gene expression in major psychiatric disorders examined DNA methylation and histone modifications in tissue homogenates, thus ignoring the fact that the gene(s)-of-interest often are expressed only in a select subpopulation of neurons or other cells.

To bypass these limitations, a recent postmortem brain study, conducted on 16 subjects on the autism spectrum and 16 controls of child and adult age, profiled the transcriptional mark, histone H3-trimethylated at lysine 4, on a genome-wide scale in prefrontal cortical neurons and separately, in non-neuronal chromatin from the same cases (Shulha et al. 2012a). The study identified 711 “epigenetic risk” loci were affected in variable subsets of autistic individuals, including the synaptic vesicle gene RIMS3, the retinoic acid signaling regulated gene RAI1 (Nakamine et al. 2008), the histone demethylase JMJD1C (Castermans et al. 2007), the astrotactin ASTN2 (Glessner et al. 2009), the adhesion molecules NRCAM (Sakurai et al. 2006; Bonora et al. 2005; Petek et al. 2001) and SEMA5 (Weiss et al. 2009; Melin et al. 2006), the ubiquitin ligase PARKIN2 (PARK2) (Scheuerle and Wilson 2011; Glessner et al. 2009) and many other genes for which rare structural DNA variations could carry high disease penetrance . Intriguingly, for most of these epigenetic changes, the H3K4me3 changes occurred selectively in prefrontal neurons, but not in their surrounding non-neuronal cells (Shulha et al. 2012a). Therefore, the disease-associated epigenetic signatures in ASD are cell-type specific and, at least in neurons, show significant overlap with the genetic risk architecture of neurodevelopmental disorders (Shulha et al. 2012a).

Another study, using peripheral blood cells from monozygotic twins discordant for ASD, identified multiple disease-associated differentially DNA methylated regions, which would suggest that at least some of the epigenetic alterations in subjects with ASD may be unrelated to the genetic risk architectures in the affected cases (Wong et al. 2013).

To date, no medication has been approved for the treatment of core symptoms of ASD, including social and other cognitive defects and problems associated with language and communication and interaction (Payakachat et al. 2012) . Preliminary data suggest that intranasal application of the nonapeptide, oxytocin , when administered in research and laboratory settings, elicits a modest improvement in repetitive behaviors and some of the test measures for social cognition (Bakermans-Kranenburg and van Ijzendoorn 2013). In addition, antipsychotic medications (originally designed to treat schizophrenia and other types of psychosis) such as risperidone and aripiprazole demonstrated improvement in parent-reported measures of challenging behaviors such as repetitive or aggressive behaviors, but carry a significant side-effect burden (Payakachat et al. 2012). Other types of drugs, including anticonvulsants and anxiolytics may be very beneficial in selected cases to treat seizures and anxiety but obviously do not address the core symptoms. Thus, there is a pressing need to pursue novel and innovative treatment options for these types of conditions. It is not unreasonable to explore promising epigenetic drug targets in ASD .

One recent, impressive example for epigenetic drug targets is provided by topoisomerases (topos), DNA cleaving enzymes that are important in the processes for replication and recombination, transcription and chromatin remodeling (Salerno et al. 2010). A recent study using an unbiased high-content screening approach, using mouse primary cortical neurons, discovered that a diverse group of molecules, that share an inhibitory activity against DNA topoisomerase type I or II enzymes can unlock the expression of the normally epigenetically silenced paternal allele of the gene encoding ubiquitin protein ligase E3A (Ube3a). These topo inhibitors mediate their effect by reducing the expression of the imprinted Ube3a antisense RNA (Ube3a-ATS) (Huang et al. 2011). The expression of this antisense RNA is normally repressed in the maternal chromosome in conjunction with the allele-specific DNA methylation of an imprinting center (a DNA or chromatin structure that carries epigenetic information about parental origin) (Bressler et al. 2001). Similar to hundreds of other loci defined by parent-of-origin-specific gene expression, Ube3a was considered epigenetically stable throughout life (Reik 2007). This hypothesis now needs to be revised, however, given that even a single intrathecal infusion of the FDA-approved topoisomerase inhibitor topotecan was sufficient to relieve silencing of the paternal Ube3a (sense) transcript in lumbar spinal neurons for an extended period of at least 3 months (Huang et al. 2011) . The most obvious explanation for topotecan’s mechanism of action—altered DNA methylation of the Ube3a imprinting center—has been ruled out, thus the underlying mechanism(s) remain a mystery. Thus, the reactivation of paternal UBE3A expression via topoisomerase inhibition could provide a starting point to investigate potential therapies for Angelman syndrome, which is caused by loss of function mutations and deletions at the maternal UBE3A locus (Kishino et al. 1997; Matsuura et al. 1997; Sutcliffe et al. 1997), and for which there are currently no effective treatments. Currently, it remains unknown whether topoisomerase-mediated reversal of imprinting-related gene expression is specific to the UBE3A locus. However, this issue could be addressed by, for example, fusing topoisomerase enzymes to customized motifs for sequence-specific binding at UBE3A. For example, zinc finger nucleases or transcription activator-like (TAL) effectors of plant pathogenic bacteria have been fused to the FokI restriction enzyme, allowing the induction of‘custom-made’ DNA strand breaks at specific and even at unique loci in the genome (Porteus and Baltimore 2003; Bogdanove and Voytas 2011). An even more promising toolkit was recently introduced to the field, utilizing a genome defense mechanism in bacteria, CRISPR (Clustered Regularly Interspaced Short Palindromic Repeats)/Cas (CRISPR-associated) (Bhaya et al. 2011). The Type II CRISPR/Cas system requires a single protein, Cas9, to catalyze DNA cleavage (Sapranauskas et al. 2011) and is now widely used to induce targeted mutations in eukaryotic systems, including human and mouse cells (Cong et al. 2013) .