Introduction

Chromatin dynamism is critical to basic cellular processes such as gene transcription, DNA replication, DNA recombination and DNA repair. DNA accessibility is modulated by epigenetic mechanisms that ultimately alter the structure of the chromatin and provide binding sites for a wide variety of regulatory proteins. The orchestrated organization of epigenetic factors, including DNA methylation, histone marks, non-coding RNAs (ncRNAs), and their associated chromatin proteins, is essential for development and cellular differentiation. For instance, extensive chromatin remodeling occurs on a global level during early development. DNA methylation patterns undergo genome-wide alterations that occur immediately after fertilization and during early-preimplantation development, together with histone modification changes, such as increased H3K9me with differentiation (Reik 2007). Epigenetic factors also guarantee the activation and maintenance of specific differentiation programs in adult somatic cells (Berdasco and Esteller 2010). The active role of epigenetic factors in controlling cellular differentiation is supported by spontaneous cell differentiation after treatment with demethylating agents or histone deacetylase inhibitors (reviewed in Berdasco and Esteller 2011). Treatment with the demethylation agent 5-aza-2′-deoxycytidine promotes differentiation of different types of adult stem cells into cardiac myogenic or osteogenic cells by enhancing the expression of lineage genes. In a similar manner, histone deacetylase inhibitor trichostatin enhances chondrogenic or neural differentiation of stem cells, reinforcing the epigenetic control of differentiation. Furthermore, the essential role of these factors is reflected in the fact that altered profiles of epigenetic marks often lead to defaults in cellular homeostasis and development of human diseases. Genetic alterations could explain the causes of several monogenic diseases. However, the genetic basis underlying the origin of complex and multifactorial diseases remains largely unknown and the importance of the role of non-genetic mechanisms, including epigenetic mechanisms or posttranslational protein modifications, is increasingly being realized. Cancer has been the best characterized complex human disease associated with epigenetic defects (Berdasco and Esteller 2010), but the list of complex diseases carrying epigenetic defaults has been increasing rapidly in recent years. Epigenetic studies have now been made of complex diseases such as obesity, type 2 diabetes mellitus, cardiovascular diseases and neurological disorders. These pathogenic mechanisms are particularly interesting because the epigenetic effects may also be affected by aspects of the environment such as diet and lifestyle, raising the possibility of “resetting” the altered epigenetic marks. Deleterious epigenetic profiles could be a consequence of mutations in the “writers”, that is to say, dysfunctional enzymes that are responsible for putting in and out the epigenetic marks. Defective epigenetic machinery has been observed in cancer initiation and progression. Furthermore, germline mutations of epigenetic modifiers contribute to the development of human diseases including intellectual disability (review in: Froyen et al. 2006; Kramer and van Bokhoven 2009; Franklin and Mansuy 2011). The aim of the present review is to provide an overview of these disorders grouped by the type of epigenetic change involved: (i) alterations in DNA methylation players; (ii) mutations in histone modifiers; (iii) disruption of chromatin-remodeling complexes, and (iv) mutations in non-coding RNA processing machinery.

Genetic disorders linked to DNA methylation defects

DNA methylation, or the addition of a methyl group to a cytosine, is a key epigenetic player that has long been considered the genome’s fifth base (Portela and Esteller 2010). In mammals this reaction occurs at CpG (deoxycytidine-phosphate-deoxyguanosine) sites located throughout the genome, but there are certain areas, known as CpG islands, that are enriched in CpG dinucleotides (especially in promoter regions). Non-CpG islands of the human genome are usually methylated and prevent genomic instability phenomena, such as the movement of transposable elements (Berdasco and Esteller 2010). Normal methylation at these sequences is also necessary for X-chromosome inactivation in females and genomic imprinting. Conversely, CpG islands are usually unmethylated, being closely related to the expression of housekeeping genes. It is estimated that only 6 % of the human CpG islands are methylated and, consequently, silenced, being essential for maintaining tissue-specific patterns during development and differentiation. By our current understanding, this “DNA methylation code” seems to be an oversimplification, since, in recent years, new genomic contexts outside of CpG islands, known as CpG shores, have emerged as candidates for regulating gene expression of tissue-specific genes. The technological advance in studying DNA methylation will provide insight into the role of 5-methylcytosine patterns with respect to their density, location and function, amongst other features. Additionally the recently discovered cytosine modification 5-hydroxymethyl-2′-deoxycytidine (5hmC) needs to be further studied to determine its implications for normal and diseased epigenetic regulation.

The enzymes responsible for introducing the methyl group into a cytosine are DNA methyltransferases (DNMTs). Three major proteins with DNMT activity have been identified in mammals: DNMT1, DNMT3A and DNMT3B. DNMT1 is a widely expressed maintenance DNMT that recognizes hemimethylated DNA and is responsible for maintaining the existing methylation patterns after DNA replication. By contrast, DNMT3 enzymes are de novo DNMTs that introduce methyl groups into previously unmethylated cytosines. These enzymes introduce a methyl group into the genome, but this “writing” must be interpreted (read) by the rest of the cellular machinery (i.e., transcription factors, DNA polymerases, chromatin-remodeling proteins, epigenetic enzymes, etc.). Additional members of the DNMT family without methyltransferase activity have been reported, such as DNMT2 or DNMT3L. DNMT3L lacks the amino acid sequence necessary for methyltransferase but it seems to be required for the establishment of maternal genomic imprints (Aapola et al. 2002).

Methyl-CpG-binding domain (MBD) proteins are one of the DNA methylation-associated proteins that could be recruited to methylated DNA and in turn facilitate the recruitment of histone modifiers and chromatin-remodeling complexes (Portela and Esteller 2010). Evidence is mounting of the role of DNA methylation in modulating cognitive functions of the central nervous system, such as learning and memory, and of how dysregulation of DNMTs activities can give rise to neurological disorders (Liu et al. 2009; Urdinguio et al. 2009; Feng et al. 2010). Some of the genetic syndromes featuring mutations in the DNA-related machinery that often cause neurological disorders are discussed in this section (Table 1).

Table 1 List of human disorders associated with germline mutations in epigenetic genes
Full size table

DNMT1 mutations and disorders of the central and peripheral nervous system

Hereditary sensory and autonomic neuropathy type 1 with dementia and hearing loss (HSAN1; MIM #614116) is a degenerative disorder of the central and peripheral nervous system. Its clinical manifestations consist of sensory impairment, sudomotor dysfunction (loss of sweating), dementia and sensorineural hearing loss. HSAN1 is inherited in an autosomal dominant manner, although the proportion of patients with de novo mutations is unknown, DNMT1 being the only gene in which mutations of exons 20 and 21 are known to cause HSAN1 (Klein et al. 2011; Winkelmann et al. 2012). Molecular genetic testing to screen for three mutations in exon 21 of DNMT1 (p.Ala570Val, p.Gly605Ala and p.Val606phe) is available for research purposes. Mutations are present within the targeting-sequence domain of DNMT1 that regulates the binding of the enzyme to chromatin during the S-phase and is responsible for maintaining this association during the G2-M phases (Fatemi et al. 2001; Song et al. 2012). DNMT1 is strongly expressed in postmitotic neurons and plays important roles in neuronal differentiation, migration and central neural connection (Feng et al. 2010). The functional involvement of DNMT1 mutations has been assessed in in vitro studies in HeLa cells (Klein et al. 2011), so that cells carrying mutations in the DNMT1 targeting sequence showed abnormal heterochromatin binding of DNMT1 during the G2 phase and were prematurely degraded. Abolishing DNMT1 function affects DNA methylation cellular levels: first, a lower level of global methylation (8 %) has been measured in mutant cells (i.e., satellite 2 methylation was reduced); second, site-specific hypermethylation at specific loci has also been found (Klein et al. 2011). Additionally, DNMT1 is required for CD4+ differentiation into T regulatory cells (Josefowicz et al. 2009), and a link between the absence of CD4+ T regulatory cells and the autoimmune response in neurological syndromes has been proposed (Winkelmann et al. 2012). The findings from several studies together suggest that DNMT1 participates in a precise mechanism of dynamic regulation of neuronal survival, but additional efforts will be needed to elucidate its pathogenic mechanisms and to explain the phenotypic variation observed between individuals bearing different mutations.

DNMT3 mutations and immunodeficiency centromeric instability facial syndrome 1

Immunodeficiency centromeric instability facial syndrome 1 (ICF1, MIM #242860) is a rare autosomal recessive disorder characterized by immune defects in association with centromere instability and facial anomalies. Several chromosomal abnormalities have been described, including the juxtacentromeric heterochromatin formation of chromosomes 1, 9 and 16, an increased frequency of somatic recombination between the arms of these chromosomes, and a marked tendency to form multibranched configurations (Ehrlich 2003). 60 % of ICF1 patients carry mutations in the de novo DNA methyltransferase DNMT3B (Xu et al. 1999; Lana et al. 2012). Mutations in the ZBTB24 gene, which encodes a transcription factor, are responsible for the ICF type 2 phenotype (de Greef et al. 2011). Hypomethylation in ICF patients commonly affect specific non-coding repetitive sequences (satellites 2 and 3, subtelomeric sequences and Alu sequences), imprinted genes and genes located in constitutive and facultative heterochromatin (Xu et al. 1999; Yehezkel et al. 2008; Brun et al. 2011), causing chromatin decondensation and chromosome instability. Recently, whole-genome bisulfite sequencing of an ICF patient harboring mutated DNMT3B and one healthy control have been performed to assess DNA methylation at base pair resolution (Heyn et al. 2012). The authors concluded that ICF patients have 42 % less global DNA methylation, especially in inactive heterochromatic regions (in accordance with previous studies). Interestingly, the methylation status of transcriptional active loci and rRNA repeats did not change, suggesting that there is a selective pressure to maintain the stability of these genomic structures (Heyn et al. 2012). In addition to methylation studies, the altered expression of more than 700 genes in ICF1 patients has been described, especially genes related to immune function, development and neurogenesis (Jin et al. 2008). Interestingly, half the upregulated genes were hypomethylated (compared with normal cells) in parallel with the loss of the histone repressive H3K27 trimethylation mark and the gain of the histone active marks H3K9 acetylation and H3K4 trimethylation (Jin et al. 2008). Not only the protein-coding genes are altered; a dramatic loss of methylation (from 80 to 30 %) was found in heterochromatic genes, which are usually aberrantly hypomethylated in cancer cells, although the hypomethylation was not always associated with their activation (Brun et al. 2011). In contrast to the dysregulation of protein-coding genes, no changes in histone marks associated with heterochromatic genes could be found (Brun et al. 2011). Finally, the genomic instability generated in ICF patients also resulted in replication defects, including shortening of the S-phase, a higher global replication fork speed and earlier replication of heterochromatic genes in S-phase (Lana et al. 2012). To conclude, the loss of DNMT3B function and the interaction of DNMT3B with histone modifications, together with the variable clinical features of the patients, make ICF samples an ideal model for investigating the epigenetic network and their molecular consequences in several biological pathways (gene transcription, DNA replication and recombination, among others).

MeCP2 genetic alterations and Rett syndrome

Rett syndrome (MIM #312750) is a progressive neurodevelopmental disorder characterized by arrested development between 6 and 18 months of age, regression of acquired skills, loss of speech, unusual stereotyped movements and intellectual disability (Zachariah and Rastegar 2012). It affects predominantly females, occurring at a frequency of 1:10,000 live births, but male patients with Rett syndrome and variable phenotype (i.e., severe to moderate congenital encephalopathy, infantile death or psychiatric manifestations) have also been described (Ravn et al. 2003; Moretti and Zoghbi 2006). Mutations in the X-linked gene encoding methyl-CpG-binding protein 2 (MeCP2), including missense, frameshift and nonsense mutations and intragenic deletions, account for the condition in up to 96 % of Rett syndrome patients (Amir et al. 1999; Moretti and Zoghbi 2006). Interestingly, the increase in MeCP2 dosage due to duplications of the locus and surrounding areas also causes the neurological disorder Lubs X-linked mental retardation syndrome (MRXSL, MIM #300260) (Van Esch et al. 2005). MeCP2 mutations in Rett syndrome patients arise in the germline although the phenotypic alterations in the neurological system appear at early postnatal stages (Zachariah and Rastegar 2012). Recently, abrogation of MeCP2 function in adult mice has been found to result in severe neurological symptoms commonly observed in Rett syndrome, such as global shrinkage of the brain, increased neuronal cell density, retraction of dendritic arbors, reduction of synaptic proteins and altered astrocytic development, among others (Nguyen et al. 2012). This is further evidence of the involvement of MeCP2 in regression from a normal mature brain to a Rett-like brain. This dynamism is of vital importance and suggests opportunities for reverting the Rett phenotype. In this regard, Rett syndrome is not a neurodegenerative disorder, neurons do not die, opening the opportunities of phenotypic reversion by means of MeCP2 restoration (Giacometti et al. 2007; Guy et al. 2007; Tropea et al. 2009).

MeCP2 protein is widely expressed in several tissues but the highest level of expression has been observed in the brain (Zachariah and Rastegar 2012). Although a higher level of MeCP2 expression has been described in neurons, especially in postmitotic neurons, deregulation of MeCP2 expression in glia cells also contributes to the progression of Rett syndrome (Ballas et al. 2009). MeCP2 was initially identified as a transcriptional repressor that binds to methylated CpG dinucleotides and recruits corepressors such as mSin3 and HDACs (Jones et al. 1998). However, in recent years, studies have concluded that MeCP2 may act either as a repressor or an activator, depending on its interaction proteins (Chahrour et al. 2008; Yasui et al. 2007). Chahrour et al. (2008) examined the gene expression profiles in the hypothalamus of mice lacking or overexpressing MeCP2 and, contrary to expectation, confirmed that MeCP2 occupancy preferentially occurs in active genes (85 %) and is associated with binding to the transcriptional activator CREB1. In accordance with this finding, similar genome-wide analyses of MeCP2 binding, CpG methylation and gene expression showed that MeCP2 binds methylated and unmethylated DNA preferentially to actively expressed genes (Yasui et al. 2007). Several MeCP2 target genes (UBE3A, DLX5, BDNF and PRODH) have been identified (Samaco et al. 2005; Horike et al. 2005; Chang et al. 2006; Urdinguio et al. 2008), although no direct link between MeCP2-dependent expression of these genes and the phenotypic abnormality of Rett syndrome has so far been found. A role for oxidative stress as a mechanism underlying the Rett phenotype has been suggested (De Felice et al. 2012) mainly on the basis of two observations: (i) oxidation of either a single guanine to 8-oxoG or of a single 5mC to 5hmC, significantly inhibits (by at least an order of magnitude) binding of MeCP2 to the oligonucleotide duplex (Valinluck et al. 2004); (ii) several MeCP2 target genes affect the oxidative stress response, such as BDNF, CREB or Prodh (Chang et al. 2006; Urdinguio et al. 2008). Finally, analysis of the posttranslational modifications of MeCP2 proteins could also provide insight into the effect on synaptic plasticity mediated by the regulation of specific genes. Recent work has shown that phosphorylation of MeCP2 at serine 421 is induced by membrane depolarization and leads to the regulation of BDNF transcription (Zhou et al. 2006). Interestingly, a mouse model showed that phosphorylation at serine 421 has widespread effects on synaptic plasticity (Li et al. 2011). Neuronal activity also influences dephosphorylation of MeCP2 at serine 80, which alters the transcription of several genes (Tao et al. 2009). Considering these results together suggests that further research is warranted into the role of these MeCP2 posttranslational modifications (and other phosphorylation sites) in neuronal plasticity. It should be stressed that recent evidence suggests that MeCP2 is not only involved in transcriptional regulation, but also possibly in RNA splicing (Young et al. 2005), chromatin condensation (similar to H1 function) (Ishibashi et al. 2008) and the silencing of repetitive elements (Muotri et al. 2010). These findings demonstrate that MeCP2 function and its involvement in Rett syndrome might be more complex than previously appreciated.

Finally, MeCP2 mutations are also linked to a broad spectrum of neurological disorders (Van Esch et al. 2005; Villard 2007), and to autism (Carney et al. 2003), Angelman syndrome (Watson et al. 2001) and Prader–Willi syndrome (Samaco et al. 2004). As mentioned before, duplications or triplications on chromosome Xq28 containing the MeCp2 region are also associated with the Lubs X-linked mental retardation syndrome (Van Esch et al. 2005). Taking together, it must be highlighted that both down- and overexpression of MeCP2 result in altered neuron function, an aspect that must be especially considered for therapeutic purposes based on MeCP2 restoration. Although it is clear that MeCP2 deficiency affects the brain function, a definitive molecular pathology of the MeCP2-associated disorders remains elusive. In this concern, progresses are currently being carried out mainly due to the development of appropriate experimental systems, such as stem cell-based system allowing the synchronous differentiation of neuronal progenitors in wild-type or mutant MeCP2 (Yazdani et al. 2010) or by the development of several mouse models that reproduce many traits of Rett syndrome (Na et al. 2012). Hopefully their findings will help us better understand the many facets of the pathobiology of the disease.

Genetic alteration of histone modifiers

The histone modification network is very complex. Post-transcriptional histone modifications can occur in various histone proteins (e.g., H2B, H3, H4) and variants (e.g., H3.3) and affect different histone residues (lysine, arginine, serine) located in their N-terminal tails (Esteller 2008; Bannister and Kouzarides 2011). Several chemical groups [methyl, acetyl, phosphate, small ubiquitin-related modifier (SUMO) and ADP-ribose] may be added in different degrees depending on the chemical modification (mono-, di- or trimethylation) (Bannister and Kouzarides 2011). Cross-talk between histone marks can occur within the same residue, in the same tail or among different histone tails (Portela and Esteller 2010) and, as a consequence, the functional significance of histone modifications depends on the combination of all marks (the “histone code”). Furthermore, we must not forget that an additional level of complexity exists due to the communication between the epigenetic marks involving DNA, histone and chromatin-related proteins. Histone modifications are involved in gene transcription, although the consequence of each mark depends on the residue affected and the type of modification. In general, acetylation of lysines is associated with transcriptional activation. However, methylation of lysine 4 of the H3 histone is associated with active transcription whereas methylation of lysines 9 and 27 is associated with gene silencing (Bannister and Kouzarides 2011). Gene transcription is not the only characteristic that is controlled in this way; histone modifications are a mechanism for controlling chromatin structure and they also affect more global biological processes such as DNA repair, DNA replication, alternative splicing and chromosome condensation (Portela and Esteller 2010).

Addition of chemical groups to histone residues is a very dynamic and reversible process catalyzed by two sets of enzymes (and their protein complexes) that have antagonistic activities, enzymes that covalently attach the chemical groups and others for removing them (Fig. 1). Acetyltransferases (HATs) and histone deacetylases (HDACs) are among the least specific histone modifiers because they are able to modify several residues. Conversely, histone methyltransferases (HMTs), histone demethylases (HDMTs) and kinases have higher specificity (Portela and Esteller 2010). Genetic alterations of histone modifier enzymes are frequently linked to human diseases. In this regard, aberrations in the histone modification profiles associated with cancer could be a consequence of the genetic disruption of the epigenetic machinery (Berdasco and Esteller 2010). Several hematological malignancies can be associated with chromosomal translocations in the coding region of HATs (i.e., CBP-MOZ) or HMTs (i.e., mixed-lineage leukemia 1 (MLL1), or nuclear receptor binding SET domain protein 1 (NSD1). In solid tumors, both HMT genes [such as EZH2, mixed-lineage leukemia 2 (MLL2)] and DNMTs [i.e., Jumonji domain-containing protein 2C (JMJD2C/GASC1)] are known to be amplified (Berdasco and Esteller 2010). In this section we will focus on non-tumoral human diseases in which the epigenetic profile changes as a consequence of genetic alterations in histone modifiers (Table 1).

Fig. 1
figure 1

Epigenetic mechanisms disrupted in human disorders. Epigenetic mechanisms regulate chromatin function and cell identity. Appropriate activity of enzymes controlling DNA methylation, histone modifications and non-coding RNAs controls the temporal and spatial patterns of gene expression, DNA repair and DNA replication. Their deregulation may contribute to human diseases. Epigenetic-based therapies, such as histone deacetylase inhibitors, can partially alter the phenotype of the disease by recovering the aberrant epigenetic patterns, and are a promising area in pharmacological research. Chemical modifications at histone H4 are shown as a representative example of histone marks. 5-mC; 5-methylcytosine; Ac histone acetylation, DNMTs DNA methyltransferases, HATs histone acetyltransferases, HDACs histone deacetylases, HDMTs histone demethylases, HMTs histone methyltransferases, MBDs methyl-binding domain proteins, Me histone methylation, P histone phosphorylation, PHD plant homeodomain, PWWP proline–tryptophan–tryptophan–proline domain, XPO5 exportin-5

Full size image

MYST4 acetyltransferase (KAT6B) mutations in Genitopatellar syndrome and Say-Barber-Biesecker-Young-Simpson syndrome

Genitopatellar syndrome (GPS, MIM #606170) is a rare skeletal dysplasia consisting of microcephaly, severe psychomotor retardation and craniofacial defects, associated with congenital flexion contractures of the lower extremities, abnormal or missing patellae, and urogenital anomalies (Reardon 2002). To date it has been described in only 18 subjects (Campeau et al. 2012a). Whole-exome sequencing identified mutations in the KAT6B acetyltransferase that lead to protein truncation (Campeau et al. 2012a; Simpson et al. 2012). All mutations are heterozygous, exhibit autosomal dominant inheritance and occur in the proximal portion of the last exon (Campeau et al. 2012a). KAT6B is a member of the MYST family of proteins containing a conserved acetyltransferase domain (Champagne et al. 1999). The absence of decreased expression of the KAT6B transcript has been described in GPS patients (Campeau et al. 2012a); however, GPS subjects were characterized by decreased global acetylation of histone H3 and H4 (Simpson et al. 2012).

Heterozygous mutations of KAT6B have also been described in patients with Say-Barber-Biesecker-Young-Simpson syndrome (SBBYS, MIM #603736) (Clayton-Smith et al. 2011). The main clinical features associated with the syndrome are distinctive facial appearance, severe hypotonia and feeding problems, associated skeletal problems, cardiac defects, severe intellectual disability, delayed motor milestones, and significantly impaired speech (Clayton-Smith et al. 2011). Unlike with GPS syndrome, mutations in SBBYS patients are located throughout the gene or more distally in the last exon. Although GPS and SBBYS share several clinical features, such as severe intellectual disability, cardiac defects and genital abnormalities, the range of phenotypic alteration varies between the two syndromes (Campeau et al. 2012b). Campeau and collaborators created a database for establishing correlations between phenotype and genotype in patients carrying KAT6B mutations. Their preliminary results suggested that the common features are a consequence of haploinsufficiency of the C-terminal region, whereas the unique phenotypes of GPS could arise from the expression of a truncated protein that acquires new cellular functions (Campeau et al. 2012b).

The molecular mechanisms of abrogation of KAT6B function that lead to defective neural development are not currently known. The Querkopf mouse model that only expressed a 5 % of the MYST4 levels quantified for wild-type mice leads to a phenotype commonly observed in human syndromes: brain development defects, facial dysmorphisms and alteration of bone growth (Thomas et al. 2000), supporting the role of KAT6B in cerebral cortex development. According to this idea, KAT6B is expressed in developing cerebral cortex, adult neural stem cells, osteoblasts and germ cells (Merson et al. 2006). A study performed in one Noonan-like phenotype patient, with a chromosomal breakpoint in KAT6B resulting in 50 % gene expression, and in Querkopf mice, described a reduction of H3 acetylation that specifically dysregulates the expression of genes in the MAPK pathway (Kraft et al. 2011).

Genetic disruption of ep300/CBP acetyltransferases in Rubinstein–Taybi syndrome

The Rubinstein–Taybi syndrome (RSTS, MIM #180849) is a well-defined disease characterized by postnatal growth deficiency, microcephaly, specific facial characteristics, broad thumbs, big toes and intellectual disability. An increased risk of tumors (mainly leukemia in childhood and meningioma in adulthood) has been observed (Hennekam 2006). Although the exact molecular etiology of RSTS is not clearly understood, it is widely accepted that RSTS is associated with breakpoints, translocations, mutations and microdeletions of chromosome 16p13.3 (Lacombe et al. 1992). Petrij et al. (1995) were the first to report that mutations in the gene encoding the CREB binding protein (CBP), located in the aforementioned region, could cause RSTS. Recent investigations employing larger series of RSTS samples detected CBP mutations in 45–55 % of patients (Roelfsema et al. 2005; Tsai et al. 2011). Furthermore, CBP has a homolog located at 22q13.2, known as the E1A-binding protein p300 (p300) (Arany et al. 1994). The exact frequency of genetic alterations of p300 in RTS is not yet known, but some sources estimate it to be 3 % (Bartholdi et al. 2007; Bartsch et al. 2010). Since a cytogenetic or molecular abnormality in p300/CBP could be detected in about 55 % of patients, further work to explain the cause of the syndrome in the remaining 45 % of patients without a “classic” genetic abnormality is still needed.

CREB binding protein was first identified as a nuclear transcription coactivator that binds specifically to CREB when it is phosphorylated (Chrivia et al. 1993), while p300 was originally described by protein-interaction assays with the adenoviral E1A oncoprotein (Eckner et al. 1994). Although both proteins are highly homologous (63 % homology at the amino acid level) and have common interaction partners, they have distinct cellular functions and cannot always replace one another (Viosca et al. 2010). There are two biological mechanisms whereby defects in p300/CBP function could cause the RTS symptoms: (i) CBP and p300 proteins act as cofactors for over 300 transcriptional factors, including several regulators involved in neuronal activity, such as c-Fos, c-Jun, CREB or NF-κß, known oncoproteins (myb), transforming viral proteins (E1A, E6, large T antigen) and tumor-suppressor proteins (p53, E2F, RB, Smads, RUNX, BRCA1) (Chan and La Thangue 2001; Kasper et al. 2011); (ii) both proteins have HAT activity that targets the N-terminal tails of histones and contributes to transcriptional activation by relaxing the structure of the nucleosomes (Ogryzko et al. 1996). CBP and ep300 have some common activities, such as the acetylation of H4K5, H3K14, H3K18, H3K27 and H3K56 (Das et al. 2009; Jin et al. 2011). However, they also have unique properties such as substrate specificity profiles that could explain functional differences of both enzymes (McManus and Hendzel 2003). RSTS disorder has been modeled in mice, and several heterozygous cbp ± mice (homozygous cbp −/− mutants are embryonic lethal) have been generated (Bourtchouladze et al. 2003; Alarcon et al. 2004). These model animals exhibit deficits in long-term memory and cognitive impairments reminiscent of human RSTS neural symptoms, confirming the role of CBP in the etiology of the disease (Alarcon et al. 2004; Korzus et al. 2004). At the molecular level, these mice have reduced HAT activity, decreased acetylation of specific histone proteins and impaired CBP-dependent gene expression (Alarcon et al. 2004; Korzus et al. 2004). The role of CBP in neural differentiation and development has been recently demonstrated in CBP genetically modified models (Wang et al. 2010). Phosphorylation of CBP by atypical protein kinase C ζ is necessary for CBP binding to neural promoters, followed by histone acetylation and transcriptional activation, leading to neural differentiation of stem cell precursors (Wang et al. 2010). This mechanism could explain how CBP alterations can result in cognitive dysfunction of RSTS patients.

HDAC4 histone deacetylase mutations in Brachydactyly-mental retardation syndrome

Brachydactyly-mental retardation syndrome (BDMR, MIM #600430) is a complex disease that presents a wide spectrum of clinical features, such as intellectual disabilities, developmental delays, sleep disturbance, craniofacial and skeletal abnormalities (including brachydactyly type E), cardiac defects and autism (Aldred et al. 2004). Heterozygous mutations in the HDAC4 deacetylase gene located on chromosome 2q37.2 have been reported in BMRS subjects (Williams et al. 2010; Morris et al. 2012). HDAC4 acts as a corepressor for transcription factors regulating the expression of genes from the osteogenic, chondrogenic, myogenic and neurogenic differentiation pathways (Miska et al. 2001; Arnold et al. 2007; Chen and Cepko 2009). HDAC4 is essential for the repression of RUNX2 and MEF2 transcription factors in normal bone development (Arnold et al. 2007). Indeed, mice with a deleted MEFC2 gene have impaired chondrogenic and osteogenic development that antagonizes the phenotype of the Hdac4 mutant mice, which is similar to the human BDMR phenotype (Arnold et al. 2007; Rajan et al. 2009). These results suggest that haploinsufficiency of HDAC4 causes BDMR through its ability to regulate important master genes of cellular differentiation.

Mutations in histone methyltransferase EHMT1 in Kleefstra syndrome

Kleefstra syndrome (MIM #610253), previously known as 9q subtelomeric deletion syndrome, is characterized by severe intellectual disability, hypotonia, brachy(micro)cephaly, epileptic seizures, flat face with hypertelorism, synophrys, anteverted nares, everted lower lip, carp mouth with macroglossia, and heart defects (Willemsen et al. 2012). The mutational landscape of Kleefstra patients includes a microdeletion in the distal long arm of chromosome 9q or intragenic loss of function mutations in the histone methyltransferase EHMT1 (Kleefstra et al. 2006, 2009), both leading to haploinsufficiency of EHMT1. EHMT1 is a specific HMT for lysine 9 at histone H3 and is involved in gene repression (i.e., the NF-kB gene) (Ogawa et al. 2002; Ea et al. 2012). It has not been established how EHMT1 disruption results in the phenotypic skills of Kleefstra syndrome, but recent research on EHMT mutants in Drosophila melanogaster demonstrates that learning and memory defects could be restored after re-expression of EHMT (Kramer et al. 2011). Interestingly, the same authors have recently identified new genetic mutations affecting epigenetic modifiers in Kleefstra syndrome subjects without mutations in the EHMT1 gene, including the methyl-binding domain MBD5, the histone methyltransferase MLL3 or the chromatin-remodeling factor SMARCB1 (Kleefstra et al. 2012). These findings highlight the crucial role of epigenetic modifiers in brain development and strongly emphasize the need to explore this area of research further.

NSD1 histone methyltransferase genetic alterations in Sotos syndrome

Sotos syndrome (MIM #117550) is an autosomal dominant condition characterized by overgrowth that results in tall stature and macrocephaly, a distinctive facial appearance, learning disability (Lapunzina 2005; Tatton-Brown and Rahman 2007), and an increased incidence of malignant neoplasms (Rahman 2005). The distinctive head shape and size has led to Sotos syndrome sometimes being called cerebral gigantism (Tatton-Brown and Rahman 2007). Most cases of NSD1 mutational mechanisms, including truncating, missense and splice-site mutations and deletions, result in loss of function of the NSD1 protein (Tatton-Brown and Rahman 2007). Mutations in the nuclear receptor SET domain containing protein-1 gene (NSD1), which encodes a histone methyltransferase of lysine residues H3K36 and H4-K20, are found in patients exhibiting the clinical symptoms of Sotos syndrome (Kurotaki et al. 2002; Rayasam et al. 2003). Recently, mutations in the NFIX gene were associated with a Sotos syndrome-like phenotype without NSD1 mutations (Yoneda et al. 2012).

NSD1 is essential for early postimplantation of embryos and NSD1 homozygous mutants are embryonic lethal, although heterozygous mutant NSD1 are viable and fertile (Rayasam et al. 2003). The NSD1 protein contains a su(var)3-9, enhancer-of-zeste, trithorax (SET) domain responsible for HMT activity and other functional domains, including plant homeodomain (PHD) and proline–tryptophan–tryptophan–proline (PWWP) domains, both of which are involved in a protein–protein interaction (Kurotaki et al. 2001). Additionally, the potential of NSD1 to mono- and dimethylate lysines K218 and K221 of the p65 subunit of the immune response gene NF-kB has been noted (Lu et al. 2010). NSD1 has previously been shown to interact with nuclear receptors, such as Nizp1, a DNA-binding transcriptional repressor (Huang et al. 1998; Nielsen et al. 2004), but to date there has been no evidence clarifying whether NSD1 mutations contribute to deregulation of brain function. It has become clear that NSD1 is a versatile protein that can act as a corepressor or coactivator, depending on the cellular context (Huang et al. 1998; Pasillas et al. 2011). According to this idea, binding of NSD1 PHD domains to target genes is guided by the presence of specific histone marks of promoters (Pasillas et al. 2011), specifically methylation at lysine H3K4 and H3K9. By binding to trimethylated H3K9, NSD1 can recognize genes that are transcriptionally repressed and interact with other repression complexes (i.e., DNMT1 or HP1), whereas its interaction with trimethylated H3K4 allows binding to active genes (Pasillas et al. 2011). In addition, patients with genetic disruption of the NSD1 gene have an increased risk of developing malignancy before adulthood, including neuroblastoma, Wilms tumors and hematological malignancies (Rahman 2005). NSD1 also has a tumor-suppressor function (Berdasco et al. 2009). NSD1 function is abrogated in neuroblastoma and glioma cells by transcriptional silencing associated with CpG island-promoter hypermethylation, and restoration of its expression demonstrates that its tumor-suppressor features are mediated by a mechanism dependent on MEIS1 expression (Berdasco et al. 2009).

Finally, occasional individuals have NSD1 defects that overlap clinically with Sotos syndrome and other conditions, such as Weaver syndrome 1 (WVS1, MIM #277590) (Douglas et al. 2003). Beckwith–Wiedemann syndrome (BWS, MIM #130650) is, like Sotos syndrome, an overgrowth syndrome. It is cause by deregulation of imprinted growth-regulatory genes within the 11p15 region. Interestingly, correlations between the two syndromes have been found: first, unexplained Beckwith–Wiedemann patients could be related to NSD1 deletions or mutations (2/52 cases) and secondly, 11p15 anomalies (including the KCNQ10T1 imprinting center) were identified in Sotos syndrome cases (2/52) (Baujat et al. 2004; Mayo et al. 2012). A potential role for NSD1 in imprinting of the 11p15 region is suggested, although the molecular basis for this association is not known.

EZH2 histone methyltransferase mutations and Weaver syndrome

Weaver syndrome 2 (WVS2, MIM #614421) is an overgrowth syndrome characterized by tall stature, advanced bone age, macrocephaly, hypertelorism, learning disabilities and dysmorphic facial features (Weaver et al. 1974). A predisposition to hematological malignancies has also been reported (Basel-Vanagaite 2010). Heterozygous mutations in the histone methyltransferase EZH2 gene on chromosome 7q36.1 have been identified in Weaver syndrome patients (Gibson et al. 2012). EZH2 protein is a member of the polycomb repressive complex 2 (PRC2), together with SUZ12 and EED, which catalyses the trimethylation of lysine H3K27 (Kirmizis et al. 2004). Mammalian EZH2 has critical roles in X-chromosome inactivations, genomic imprinting during germline development, stem cell maintenance and cell lineage determination, including osteogenesis, myogenesis and hematogenesis (Chou et al. 2011; Wyngaarden et al. 2011). A role for EZH2 in regulating the circadian-clock functions has been suggested (Etchegaray et al. 2006). In addition, mice with targeted mutant EZH2 beta cells have reduced beta cell proliferation and beta cell mass (Chen et al. 2009), whereas mice with EZH2 mutant satellite cells exhibit defects in muscle regeneration (Juan et al. 2011). Some characteristics of these phenotypes are shared with that of human Weaver syndrome, such as defects in limb development (Wyngaarden et al. 2011). However, to date, few studies have provided any insight into the specific contribution of EZH2 mutations in the Weaver phenotype. As described above, some patients with Weaver syndrome have a mutated NSD1 gene that is responsible for the overgrowth Sotos syndrome (Douglas et al. 2003). There are some clinical features common to both syndromes (i.e., developmental delay, overgrowth and macrocephaly), but some features are specific to Weaver syndrome 2 (i.e., retrognathia with a prominent chin crease or carpal bone age) (Gibson et al. 2012). Indeed, EZH2 protein acts in the PI3K/mTOR pathway, which has been associated with growth defects. This suggests that the pathways through NSD1 and EZH2 HTMs that contribute to overgrowth disorders can be different (Tatton-Brown et al. 2011).

Histone methyltransferase (MLL2) and demethylase (JMJD3) mutations in Kabuki syndrome

Kabuki syndrome 1 (MIM #147920) is an autosomal dominant intellectual disability syndrome with additional features, including a highly distinctive recognizable facial phenotype characterized by long palpebral fissures with eversion of the lateral third of the lower eyelids, a broad and depressed nasal tip, large prominent earlobes, scoliosis, short fifth finger, persistence of fingerpads, radiographic abnormalities of the vertebrae, hands, and hip joints, and others (Niikawa et al. 1981). Mutations in the histone methyltransferase MLL2 gene are a major cause of Kabuki syndrome (type 1) (Ng et al. 2010; Hannibal et al. 2011), but mutations in the histone demethylase KDM6A have also been found in Kabuki syndrome (type 2, MIM #300867) (Lederer et al. 2012). MLL2 is a lysine H3K4-specific histone methyltransferase that belongs to the SET1 family of proteins (Dillon et al. 2005), whereas KDM6A (JMJD3) is a histone demethylase that specifically acts in mono- di- and trimethylated lysine H3K27 (Hong et al. 2007; Lan et al. 2007). Both enzymes help regulate genes from the myogenic lineage during embryogenesis (Aziz et al. 2010; Seenundun et al. 2010). The identification of mutations in MLL2 and KDM6A suggests a crucial effect of altered histone methylation profiles on the phenotype of Kabuki syndrome.

Histone demethylase PHF8 mutations in Siderius X-linked mental retardation syndrome

Siderius X-linked mental retardation syndrome (MRXSSD, MIM #300263) is an inherited condition that was first described in 1999 as a heterogeneous intellectual disability syndrome associated with cleft lip/cleft palate (Siderius et al. 1999). Mutations in the PHD finger protein 8 (PHF8) gene located on the Xp11 chromosome have been identified in subjects with MRXSSD (Laumonnier et al. 2005; Abidi et al. 2007; Koivisto et al. 2007). The PHF8 protein contains a PHD-type domain zinc finger domain and a Jumanji domain, the latter conferring histone demethylating catalytic activity with specificity for: mono- and dimethylated histone H3 at lysine K9 (Yu et al. 2010) and mono-methyl histone H4 at lysine K20 (Qi et al. 2010). PHF8 transcript is strongly expressed in the embryonic and early postnatal steps of brain development (Laumonnier et al. 2005), implying a connection between PHF8 mutations and the MRXSSD phenotype. Some evidences of the functional role of PHF8 in different experimental models exist (Liu et al. 2010; Qi et al. 2010). In zebrafish, it has been described that PHF8 regulates brain apoptosis and craniofacial development through the transcriptional gene regulation (Qi et al. 2010); in Caenorhabditis elegans, a RNA interference-based functional genomic project identified PHF8 as a gene involved in controlling cellular growth and differentiation during embyogenesis (Fernandez et al. 2005); in human HeLa cells, PHF8 deficiency by siRNA mechanisms leads to a delay in G1/S transition and its dissociation from chromatin in early mitosis which demonstrates an active role of PHF8 in the control of cell cycle (Liu et al. 2010). Interestingly, PHF8 interacts with another MRXSSD protein, the transcription factor ZNF711 (Kleine-Kohlbrecher et al. 2010). PHF8 and ZNF11 proteins share a set of target genes, being of special relevance the interaction of both proteins with JARID1C (KDM5C) (also involved in alterations of the intelectual ability) (Kleine-Kohlbrecher et al. 2010; Claes et al. 2000). In addition to MRXSSD syndrome, defects affecting the chromosomal region of PHF8 (a larger Xp11.22 deletion that includes the FAM120C and WNK3 genes) have also been associated with autism (Qiao et al. 2008).

Histone demethylase JARID1C (KDM5C) mutations in Claes-Jensen X-linked mental retardation syndrome

Mutations in the histone demethylase JARID1C (KDM5C) were first identified in individuals with X-linked mental retardation syndrome with additional features of progressive spastic paraplegia, facial hypotonia, aggressive behavior and strabismus (MIM #300534) (Claes et al. 2000; Jensen et al. 2005). Heterogeneous clinical features associated with XLMR and mutated KDM5C have been consistently identified since then (Abidi et al. 2008; Santos-Rebouças et al. 2011; Ounap et al. 2012). KDM5C gene encodes a specific histone H3 demethylase at lysine K4 (Tahiliani et al. 2007). It is ubiquitously expressed although fetal brain tissues have higher KDM5C expression levels than other tissues (Xu et al. 2008). The transcriptional repressor activity of KDM5C is mediated by the Re-1 silencing transcription factor (REST) complex (Tahiliani et al. 2007). Interestingly, knock-out of the KDM5C complex results in increased trimethylation at H3K4 and gain of expression of SCN2A and SYN1 neurological genes, and provides evidence of the contribution of KDM5C complex to X-linked mental retardation defects (Tahiliani et al. 2007).

Mutations in chromatin remodelers

Nucleosome positioning and, consequently, DNA accessibility may be controlled by mechanisms that are independent of histone-modifying enzymes. Several groups of protein complexes (“chromatin-remodeling complexes”) are known to restructure nucleosomes in an ATP hydrolysis-dependent manner. To date, four families of chromatin remodelers have been described in eukaryotes: SWI/SNF, ISWI, NURD/Mi-2/CHD, and INO80/SWR1 (Hargreaves and Crabtree 2011). The ATPase domain is a common feature, but the composition of the different subunits comprising the complex is highly variable. In a similar manner, each ATPase domain may be targeted to specific domains (i.e., bromodomain, DNA helicase, etc.). Together, the binding affinity and the complex composition confer unique features on chromatin remodelers in a wide range of biological processes and genomic contexts.

SWI/SNF is one of the best studied chromatin-remodeling complexes in human cells and is composed of at least 15–20 subunits, including ATPases, catalytic subunits (i.e., SMARCA2 and SMARCA4) and structural components involved in target recognition or stabilization functions (i.e., ARID1A, ARID1B and SMARCC1) (Hargreaves and Crabtree 2011). All members of this complex contain either SMARCA2 (also known as Brahma protein) or SMARCA4 (also known as BRG1) as a catalytic unit. The two proteins share 75 % amino acid homology (Santen et al. 2012). The SWI/SNF complex plays a crucial role in cell differentiation (Ho et al. 2009), cell cycle (Nagl et al. 2007) and DNA repair (Park et al. 2006). In the human ISWI (imitation switch) family of chromatin remodelers, the catalytic subunit is represented by SNF2H and SNF2L proteins (Flaus and Owen-Hughes 2011). ISWI complexes participate in biological functions such as chromatin assembly, nucleosome spacing, DNA replication and activation or repression of transcriptional regulation (Erdel and Rippe 2011). The CHD family also contains a SNF2L ATPase domain together with tandem chromodomains in the N-terminal region (Murawska and Brehm 2011). CHD complexes are highly versatile and although they are involved in transcriptional regulation, various CHD regulatory complexes are involved in the initiation, elongation or termination of transcription. Furthermore, specific CHD complexes, such as Mi-2/nuRD, contain HDAC and MBD proteins in the same complex (Murawska and Brehm 2011). The INO80 subfamily is the most recently identified SWI/SNF family of chromatin remodelers. Mammalian INO80 complex comprises the INO80 catalytic unit and Snf2-related CBP activator protein (SRCAP) and p400 subunits (Morrison and Shen 2009). The INO80 subfamily is the most evolutionarily conserved of all the chromatin-remodeling complexes due to the high degree of homology of its ATPase subunit (Morrison and Shen 2009). Apart from regulation of transcription, the INO80 complex is involved in genome stability pathways, such as DNA repair, replication, telomere regulation and centromere stability (Ho and Crabtree 2010). In summary, ATP-dependent enzymes that remodel chromatin are important regulators of chromatin dynamism. Evidence is emerging that alterations in such chromatin-remodeling complexes have consequences for normal development. Some examples of genetic mutations of chromatin-remodeling complexes in human diseases are summarized in this section (Table 1).

ATXN7 mutation in Spinocerebellar Ataxia 7

Spinocerebellar Ataxia Type 7 (SCA7; MIM 164500) is an autosomal dominant inherited neurodegenerative disorder characterized by progressive cerebellar ataxia, including dysarthria and dysphagia, and cone-rod and retinal dystrophy with progressive central visual loss resulting in blindness in affected adults. The disease is caused by an expanded CAG trinucleotide repeat encoding a polyglutamine tract in ataxin-7 (ATXN7) gene, from 4 to 35 repeats in normal gene to a variable expansion of 36–306 repeats in pathogenic ATXN7 variants (Garden and La Spada 2008). ATXN7 protein is a transcription factor with important roles in chromatin regulation through its effect on histone modification and histone deubiquitination. It is a member of the transcription coactivator complex STAGA (SPT3/TAF9/GCN5) with acetyltransferase activity, but may also be found in the USP22 deubiquitination complex (Sopher et al. 2011). Although the nuclear expression of ATXN7 is necessary for transcriptional regulation, a functional role for cytoplasmic ATXN7 in the regulation of cytoskeletal dynamics (mediated by its interaction with microtubules) has recently been proposed (Nakamura et al. 2012). Defects of the nuclear ATXN7 gene have been correlated with the SCA7 phenotype (Chen et al. 2012a, b), although the molecular mechanisms underlying the disease are not clearly understood and the effect of epigenetic dysregulation of target genes is still a matter of debate. Some results from yeast and mice indicate that loss of the Gcn5 acetyltransferase function triggered by polyQ-Atxn7, resulting in chromatin structure changes, could be involved in the SCA7 phenotype (Yoo et al. 2003; McMahon et al. 2005; Helmlinger et al. 2006). In contrast, loss of Gcn5 functions in mice bearing polyQ-Atxn7 accelerates neuronal dysfunction in a mechanism that is independent of gene expression changes (Chen et al. 2012a, b). Deciphering the exact causal consequences of ATXN7 dysregulation in SCA7 disease through its role as nuclear transcription regulator or cytoplasmic function will need further research.

ATRX mutations in alpha-thalassemia X-linked mental retardation syndrome

ATR-X syndrome (MIM #301040) is an X-linked disorder comprising severe psychomotor retardation, characteristic facial features, genital abnormalities, and the blood disease alpha-thalassemia (Gibbons et al. 1995). Mutations in the ATRX gene located at Xq21.1 and coding for a member of the SWI/SNF chromatin-remodeling family of proteins underpin the molecular genetics of the disease (Gibbons et al. 1995). The X-linked mental retardation-hypotonic facies syndrome (MIM #309580) and the alpha-thalassemia myelodysplasia syndrome (MIM #300448) are also associated with mutations in the ATRX gene (Abidi et al. 2005; Gibbons et al. 2003). The N-terminus contains a globular domain, called ADD (ATRX-DNMT3-DNMT3L) that can bind to the N-terminal of histone H3 (Argentaro et al. 2007). Indeed, ATRX is known to be required for the incorporation of the histone H3.3 specifically at telomeric sequences (Lewis et al. 2010). On the other hand, the C-terminus contains seven helicase/ATPase domains that share sequence homology with the SNF2 family of proteins (Picketts et al. 1996). Through this domain, ATRX shows in vitro ATP-dependent nucleosome remodeling and DNA translocase activities (Gibbons et al. 2003). The ATRX protein is expressed genome-wide, but is enriched at telomeric and subtelomeric regions (Law et al. 2010). In this context, decreased ATRX expression is associated with altered expression of telomere-associated RNA (Goldberg et al. 2010) and the DNA-damage response during S-phase at telomeric regions of pluripotent stem cells (Wong et al. 2010). The mechanisms by which ATRX is recruited to telomeric regions are not fully understood, although ATRX binding depends on trimethylation at K9H3 (Kourmouli et al. 2005). It binds to tandem repeat sequences with G-rich motifs and has been predicted to form non-B DNA structures (Law et al. 2010). More importantly, the size of the tandem repeats located in specific genes influences their expression (Law et al. 2010), providing a molecular explanation of how the same mutation at the ATRX gene can result in different phenotypes. Furthermore, ATRX mutations have been correlated with alterations in the DNA methylation patterns of highly repeated sequences, including rDNA, Y-specific satellite and subtelomeric repeats (Gibbons et al. 2000). The interplay between ATRX and the DNA methylation machinery is reinforced by the discovery that the methyl-binding domain protein MECP2 targeted the C-terminal helicase domain of ATRX to heterochromatic foci (Nan et al. 2007). MeCP2 is also mutated in Rett syndrome, so the finding suggests that alteration of the MECP2–ATRX interaction leads to pathological changes that contribute to the intellectual disability phenotype observed in both syndromes.

Mutations in ERCC6 and Cockayne syndrome

Cockayne syndrome types A (CSA; MIM #216400) and B (CSB; MIM #133540) are autosomal recessive disorders caused by mutation in the ERCC8 and ERCC6 genes, respectively (Henning et al. 1995; Laugel et al. 2010). Approximately 62 % patients diagnosed with Cockayne syndrome carry mutations of the ERCC6 gene (Laugel et al. 2010). The syndromes are characterized by severe postnatal growth failure, progressive neurological dysfunction and traits reminiscent of normal aging, such as visual impairment and sensorineural hearing loss and loss of adipose tissue (Licht et al. 2003). ERCC (excision repair cross-complementing) genes are part of the nucleotide excision repair (NER) pathway, which are responsible for removing DNA lesions such as UV-induced DNA damage. ERCC6 is a nuclear protein containing a SWI/SNF-like ATPase domain, a nucleotide-binding domain and an ubiquitin-binding domain (Anindya et al. 2010). Apart from NER functions, ERCC6 is also involved in transcription regulation, chromatin maintenance and remodeling (Newman et al. 2006). At the transcriptional level, CSB cooperates with the NurD/CHD4 complex for controlling transcription of rRNA genes (Xie et al. 2012). In this regard, CHD4/NuRD is involved in maintaining silenced rRNA genes but in permissive contexts (“poised” for transcription), whereas CSB mediates the transition from the permissive to the active state (Xie et al. 2012). CSB-mediated activation could be due, at least in part, in conjunction with the CSB–G9a interaction, to an increase in trimethylated K9H3 and recruitment of Pol-I to chromatin (Yuan et al. 2007). A new chromatin connection for CSB has recently been proposed (Batenburg et al. 2012). Primary fibroblasts derived from a CSB patient had a dysfunctional telomere structure (Batenburg et al. 2012). CSB knockdown was accomplished with alterations in TERRA, a large non-coding telomere repeat-containing RNA, resulting in alterations of telomere length and integrity (Batenburg et al. 2012). Finally, a role for CSB in controlling key mitochondrial functions in addition to the nucleolus function has been proposed (Berquist et al. 2012).

SRCAP mutations and Floating-Harbor syndrome

Floating-Harbor syndrome (FHS; MIM #136140) is a rare condition characterized by proportionate short stature, delayed osseous maturation, language deficits and a typical facial appearance. Mutations in the SNF2-related CBP activator protein (SRCAP) cause the FHS syndrome (Hood et al. 2012). SRCAP expression has also been linked to cancer, whereby it positively modulates PSA antigen expression and promotes proliferation in prostate cancer cells (Slupianek et al. 2010) and potentiates Notch-dependent gene activation (Eissenberg et al. 2005). With regard to its molecular activity, SCARP catalyzes in vitro incorporation of the histone variant H2A.Z into chromatin (Ruhl et al. 2006), a histone with a well-known function in transcription regulation and cell-cycle progression. As an example, SRCAP expression in yeast is important for the deposition of such histone variants in specific promoters like SP-1, G3BP and FAD synthetase (Wong et al. 2007). Interestingly, the demethylation effect after 5-Aza-2′-deoxycytidine treatment, a drug approved by the US Food and Drug Administration (FDA) for the treatment of hematological malignancies, requires the activity of SCARP to introduce H2A.Z, which facilitates the acquisition of nucleosome-free regions (Yang et al. 2012). On the other hand, SRCAP is also an interaction partner of the histone acetyltransferase CBP, meaning that SRCAP-CBP colocalization may occur at transcriptionally active sites (Monroy et al. 2001).

CHD7 mutations and CHARGE syndrome

The acronym CHARGE (MIM # 214800) stands for coloboma of the eye, heart anomaly, choanal atresia, retardation of mental and somatic development, genital and/or urinary abnormalities and ear abnormalities and/or deafness (Sanlaville et al. 2006). It is an autosomal dominant condition with genotypic heterogeneity, although most cases are due to the mutation or deletion of the chromodomain helicase DNA-binding domain protein-7 (CHD7), a member of SNF2-like ATP-dependent chromatin-remodeling enzymes (Sanlaville et al. 2006). CHD7 mutations have been also identified in Kallmann syndrome, a developmental disorder that shares with CHARGE some phenotypic features such as impaired olfaction and hypogonadism (Kim et al. 2008). Mice with heterozygous mutations in CHD7 are a good model for studying CHARGE syndrome, and analyses of mouse mutant phenotypes have demonstrated a role in the development and function of the neuronal system. CHD7 is necessary for mammalian olfactory tissue development and function (Layman et al. 2009), proliferation of inner ear neuroblasts and inner ear morphogenesis in mice (Hurd et al. 2010), promotion of the formation of multipotent migratory neural crest that gives rise to craniofacial bones and cartilages, and the peripheral nervous system, amongst others (Bajpai et al. 2010). Recent in vitro studies have suggested that CHD7 may directly regulate BMP4 expression, a protein involved in cartilage and bone formation, by binding with an enhancer element downstream of the BMP4 locus (Jiang et al. 2012). More CHD7 targets have been identified, such as the CHD7-dependent regulation (in association with BRG1) of SOX9 and TEIST1 genes in human neural crest cells (Bajpai et al. 2010). Mechanisms in which CHD7 regulates downstream genes vary in a tissue- and cell-specific manner and depend on specific binding to methylated histone H3 lysine 4 in enhancer regions (Schnetz et al. 2009). Although further research is needed, all these findings suggest that mutations in CHD7 could transcriptionally deregulate tissue-specific genes and developmental genes resulting in the CHARGE phenotype.

Mutations in SWI/SNF complex family genes in Coffin-Siris syndrome and mental retardation

Coffin-Siris syndrome (CSS, MIM #135900) or “fifth digit” syndrome is a multiple congenital anomaly-mental retardation syndrome characterized by severe developmental delay, coarse facial features, hirsutism and absent fifth fingernails, toenails and distal phalanges (Santen et al. 2012). In a recent study, 87 % of patients with CSS carried a mutation in one or more members of the SWI/SNF family of genes, which includes SMARCB1, SMARCA4, SMARCA2, SMARCE1, ARID1A and ARID1B (Santen et al. 2012; Tsurusaki et al. 2012). Interestingly, CSS patients carrying different genetic mutations of the SWI/SNF chromatin-remodeling factors gave rise to similar CSS phenotypes (Tsurusaki et al. 2012), suggesting a general role for these complexes in coordinating chromatin conformation and gene expression. Deregulation of the SWI/SNF complexes is also a common feature of tumorigenesis through its function in mammalian differentiation, proliferation and DNA repair (Reisman et al. 2009). However, the link between SWI/SNF mutations and intellectual disorders is still unclear. SMARCA2 and SMARCA4 are catalytic subunits with ATPase activity, while ARID1A and ARID1B are structural subunits involved in target recognition and protein–protein interactions (Hargreaves and Crabtree 2011). Both types of subunit are necessary to regulate the transcription of several genes, such as c-FOS, vimentin, CD44, cyclins, E-cadherin and important transcription factors that have been functionally linked to SWI/SNF (Reisman et al. 2009; Santen et al. 2012).

Mutations in SMARCA2 have also been described in patients with the Nicolaides–Baraitser syndrome (NBS, MIM # 601358), which is characterized by severe intellectual disability, early-onset seizures, short stature, dysmorphic facial features and sparse hair (Van Houdt et al. 2012; Wolff et al. 2012). Interestingly, Harikrishnan et al. (2005) found that SMARCA2 associates with MECP2 and regulates FMR1 gene repression in mouse fibroblasts and human T-lymphoblastic leukemia cells. Methylation at promoter sites specified the recruitment of MECP2/SMARCA2; while inhibition of methylation was associated with complex release (Harikrishnan et al. 2005). These results highlight an interesting link between epigenetic marks and ATPase-dependent chromatin remodeling. How SWI/SNF deregulation produces altered expression patterns of genes associated with the CSS or NBS phenotype is not clear, although some clues about the role of the complex in neural differentiations could help interpret some features (Seo et al. 2005; Lessard et al. 2007).

Mutations in non-coding RNA machinery

Non-coding RNAs, defined as functional RNA molecules that are not translated into a protein, may also contribute to the genesis of many human disorders (Table 1) (Esteller 2011). The best characterized ncRNAs in human conditions are microRNAs (miRNAs) (Croce 2009), although other ncRNAs members are emerging, such as small nucleolar RNAs (snoRNAs), PIWI-interacting RNAs (piRNAs), large intergenic non-coding RNAs (lincRNAs), long non-coding RNAs (lncRNAs) and transcribed ultraconserved regions (T-UCRs), among others. If we focus on miRNAs biogenesis, miRNAs are transcribed as individual units (named primary miRNA (pri-miRNA)). After processing by the Drosha complex, precursor miRNAs (pre-miRNAs) are exported from the nucleus by the protein exportin 5 (XPO5). Further processing by Dicer and TAR RNA-binding protein 2 (TARBP2) generates mature miRNAs, which are included into the RNA-induced silencing complex (RISC). Once in this complex, miRNAs could exert their function through degradation of protein-coding transcripts or by translational repression. ncRNAs profiles are frequently disrupted in different types of cancer and in non-tumoral disorders, such as imprinting disorders, rheumatoid arthritis, Rett syndrome and Alzheimer’s disease (Esteller 2011). Widespread alterations of ncRNAs profiles could be also a consequence of genetic mutations of the ncRNA-associated machinery. Some recent examples in this area are discussed in this section.

TARDBP mutations and amyotrophic lateral sclerosis 10 with or without frontotemporal lobar degeneration

Amyotrophic lateral sclerosis 10 (ALS10, MIM #612069) with or without frontotemporal lobar degeneration (FTLD) is an autosomal dominant neurodegenerative disorder characterized by death of the motor neurons in the brain, brainstem and spinal cord, resulting in fatal paralysis and respiratory failure with a typical disease course of 1–5 years. Heterozygous mutations in the TAR DNA-binding protein 43 (TARDBP) on chromosome 1p36, encoding for the TDP43 protein, are present in a 50 % of individuals affected by ALS10 (Ling et al. 2010). TARDBP is a member of the miRNA machinery. It has a double effect on miRNA pathway. First, it enhances precursor miRNA (pre-miRNA) production by both interacting with the nuclear complex of Drosha or by direct binding to the primary miRNAs (pri-miRNAs) in the nucleus; and secondly, it binds to the terminal loops of pre-miRNAs in the cytoplasm by interaction with the Dicer complex (Kawahara and Mieda-Sato 2012). Under non-pathological conditions, TARDBP is mainly localized inside the nucleus, but altered cellular distributions, including neuronal cytoplasmic, intranuclear inclusions and dystrophic neurites or glial cytoplasmic locations are found in ALS10 and FTLD (Arai et al. 2006; Neumann et al. 2006; Lagier-Tourenne et al. 2010). Mutant TARDBP mice models developed a similar phenotype than human TARDBP mutation (Wegorzewska et al. 2009). Interestingly, no cytoplasmic aggregates were found in mice mutants; suggesting that other mechanisms rather than toxic cytoplasmic aggregation are underlying the molecular basis of ALS degeneration (Wegorzewska et al. 2009). Results are in accordance with a recent paper in which TDP-43 depletion in differentiated Neruo2a results in decreased expression of miR-132-3p and miR-132-5p. Further research of the specific involvement of the aforementioned miRNAS (and others to be explored) will strongly contribute to the understanding of the pathogenesis of ALS10.

DGCR8 mutations and DiGeorge syndrome

DiGeorge syndrome (MIM #188400) is a complex disorder characterized by learning disabilities, characteristic facial appearance, submucous cleft palate, conotruncal heart defects, thymic hypoplasia or aplasia, neonatal hypocalcemia, psychiatric illness and susceptibility to infection due to a deficit of T cells (Goodship et al. 1998; Shiohama et al. 2003). DiGeorge syndrome is caused by a 1.5 to 3.0-Mb hemizygous deletion of chromosome 22q11.2 comprising the DiGeorge syndrome critical region gene 8 (DGCR8) (Shiohama et al. 2003) that encodes a double stranded RNA-binding protein that is essential for miRNA biogenesis. Specifically, DGCR8 is required in miRNA maturation for processing pri-miRNAs to release pre-miRNAs in the nucleus (Han et al. 2006). Genetic modified mouse models carrying a hemizygous chromosomal deficiency on chromosome 16 that spans a segment syntenic to the 1.5-Mb 22q11.2 microdeletion showed alterations in the biogenesis of a set of miRNAs in the brain (Stark et al. 2008). Furthermore, DGCR8 deficiency resulted in alterations of dendritic morphology, impaired sensorimotor gatin and memory alterations similar to human DiGeorge phenotype (Stark et al. 2008). Additionally, it has been also described that inactivation of a Dgcr8 conditional allele in neural crest cells results in cardiovascular defects (Chapnik et al. 2012). Similar results have been found in DGCR8 conditional knock-out mice embryos and knockout vascular smooth muscle cells (Chen et al. 2012a, b). DGCR8 deficiency was associated with down-regulation of the miR-17/92 and miR-143/145 clusters in vascular smooth muscle cells, reduced cell proliferation and increased apoptosis (Chen et al. 2012a, b). These data provide specific explanations for cardiovascular and neuronal defects that could explain, at least in part, the DiGeorge syndrome phenotype.

DICER mutations in multinodular 1 Goiter disease

Autosomal dominant multinodular Goiter (MNG, MIM #138800) is a disorder characterized by nodular overgrowth of the thyroid gland. In MNG type 1, some females may also develop Sertoli–Leydig ovary tumors (Rio Frio et al. 2011). Heterozygous mutations in DICER, a gene encoding an RNase III endonuclease essential for microRNA processing, have recently been linked to MNG pathogenesis (Rio Frio et al. 2011). Mutations in DICER are also associated with pleuropulmonary blastoma (Hill et al. 2009) and play a critical role in normal cardiac function (Chen et al. 2008). DICER contains two RNase III domains and a PAZ domain, a module that binds the end of double-strand RNA (Macrae et al. 2006). Familial MNG shows clear selective disruption of the PAZ domain, suggesting a potential role of this domain in thyroid development (Rio Frio et al. 2011). At the functional level, lymphoblasts taken from MNG patients showed altered miRNA compared with control profiles (i.e., LET7A and miR345), suggesting a dysregulation of gene expression patterns (Rio Frio et al. 2011). Further determination of the consequences of specific microRNAs synthesis could be an important topic for future research into MNG and other DICER-associated pathologies.

Therapeutic applications of epigenetics

One of the main characteristics of epigenetic mechanisms is their reversibility, making them potentially powerful tools for curative pharmacological therapy (Fig. 1). Reactivation of epigenetically silenced genes has been possible for years by the treatment with DNA demethylation drugs, such as zebularine or 5-aza-2′-deoxycytidine (5-ADC), or with histone deacetylase (HDAC) inhibitors, including SAHA (suberoylanilide hydroxamic acid), valproic acid (VPA) and trichostatin A (TSA). Indeed, some of these drugs have significant antitumoral activity and the FDA has approved the use of several of them for treating patients (Kaminskas et al. 2005; Fiskus et al. 2008; Scuto et al. 2008). This approval has sparked a dramatic increase in the development and trials of “epigenetic drugs” for treating cancer and neural diseases (Kazantsev and Thompson 2008; Heyn and Esteller 2012). Although the most advanced clinical trials are those corresponding to cancer treatment, interest has grown in the fields of neurological and neurodegenerative diseases in recent years (Day and Sweatt 2012). The possibilities have only just begun to be explored in human patients, but the basis of this therapy has been confirmed in animal models.

Rubinstein–Taybi syndrome (RSTS) is probably the best model for studying the therapeutic uses of HDAC inhibitors that could compensate for the deficiency of HAT activity (CBP mutations). RSTS has been modeled in mice, and several heterozygous cbp ± mice (homozygous cbp −/− mutants are embryonic lethal) have been generated (Alarcon et al. 2004). These models feature deficits in long-term memory and cognitive impairments reminiscent of human RTS neural symptoms, confirming the role of CBP in the etiology of the disease. At the molecular level, these mice have reduced HAT activity, decreased acetylation of specific histone proteins and impaired CBP-dependent gene expression (Alarcon et al. 2004). Treatment with HDAC inhibitors, such as SAHA or TSA, ameliorate deficits in synaptic plasticity and cognition in cbp ± mice (Hallam and Bourtchouladze 2006; Vecsey et al. 2007) by enhancing transcriptional expression of specific neuronal genes. In a similar manner, immortalized human lymphocytes derived from patients with RSTS showed reduced acetylation levels, which primarily affect histones H2A and H2B, compared to the histone acetylation levels of immortalized human lymphocytes derived from patients with Cornelia de Lange syndrome (a neurological disorder) or healthy controls (Lopez-Atalaya et al. 2012). Interestingly, the acetylation deficits in RSTS cells were rescued by treatment with TSA (Lopez-Atalaya et al. 2012).

Some therapeutic approaches have been investigated in Rett syndrome. Mecp2 knock-out mice also exhibit the neurodevelopmental phenotype characteristic of human Rett syndrome (Shahbazian et al. 2002). The cognitive defect of MeCp2-deficient mice can be reverted by MecP2-induced overexpression in mice (Collins et al. 2004) and in MeCp2-deficient astrocytes (Lioy et al. 2011), suggesting that this might well be an effective treatment for Rett syndrome. Apart from this gene therapy strategy, which is not really applicable in humans, pharmacological treatments based on epigenetic targets are beginning to be explored as more feasible therapeutic interventions. Since MeCP2 binds directly to methylated promoters in association with the corepressor complex Sin3 and HDAC, another therapeutic strategy could be based on targeting HDAC activity. It has been widely demonstrated that HDAC inhibitors enhance memory formation and neuronal postnatal formation in various experimental models (Kazantsev and Thompson 2008). There are some examples of the benefits of such treatments in the literature: the alleviation of motor deficits in mouse models of Huntington disease by injections of SAHA (Hockly et al. 2003); the delay of neurodegeneration and tauopathy in a mouse model of Alzheimer’s disease through the activation of SIRT1 by means of resveratrol treatments (Kilgore et al. 2010); and the reactivation of the frataxin genes by treatments with benzamide-based HDAC inhibitors in immortalized cultured cells from Friedrich’s Ataxia syndrome patients (Herman et al. 2006).

However, many questions remain unsolved concerning drug potency, selectivity and permeability. Treatment with HDAC inhibitors results in the re-expression of a limited number of memory-associated genes (Vecsey et al. 2007). This raises a question: are HDAC inhibitors’ treatments potentially able to restore all types of genes? Or by contrast do more “easier reverted” genes exist that can be pharmacologically manipulated? If reversion is conditioned to any factor (i.e., their genomic context or the acetylation level) HDAC treatments would be more effective in certain diseases than others. Multiple epigenetic enzymes could contribute to a specific mark (especially histone acetylation) and furthermore, manipulating histone modifications could also affect DNA methylation. Which is the best target for inducing specific gene re-expression? Should systemic or enzyme-targeted drugs be employed? This lack of specificity means that almost none of the available epigenetic drugs are isoform-specific and, for instance, HDAC inhibitors have binding affinities for different HDACs isoforms (Kilgore et al. 2010). It is expected that development of subclass-specific inhibitors might contribute to minimize the side effects of the epigenetic therapy. On the other hand, the tissue also influences its effectiveness. DNA demethylating drugs are deoxycytidine analogs that must be incorporated into the DNA after replication cycles. As a consequence, the “speed” of base substitution is strongly influenced by the rate of cell division. It is to be expected that such DNA demethylating treatments will not be very effective in neuronal cells with few or no cell divisions. Further knowledge about the pathogenesis and identification of the molecular/regulatory pathways that are altered in the disease will enable the optimal target to be identified and the therapeutic potential of epigenetic-based drugs to be exploited. Despite these difficulties, epigenetic-based therapy may become a successful intervention in the treatment of human disorders associated with epigenetic alterations.