Neurodevelopmental disorders (NDDs), including intellectual disability (ID) and autism spectrum disorders (ASD), are a large group of disorders in which early insults during brain development result in a wide and heterogeneous spectrum of clinical diagnoses. Mutations in genes coding for chromatin remodelers are overrepresented in NDD cohorts, pointing towards epigenetics as a convergent pathogenic pathway between these disorders. In this review we detail the role of NDD-associated chromatin remodelers during the developmental continuum of progenitor expansion, differentiation, cell-type specification, migration and maturation. We discuss how defects in chromatin remodelling during these early developmental time points compound over time and result in impaired brain circuit establishment. In particular, we focus on their role in the three largest cell populations: glutamatergic neurons, GABAergic neurons, and glia cells. An in-depth understanding of the spatiotemporal role of chromatin remodelers during neurodevelopment can contribute to the identification of molecular targets for treatment strategies.
A mature brain is the product of its development. Early developmental insults during the assembly of these neuronal circuits can severely impact how a person develops and behaves in their adult life. Across the ongoing developmental continuum of progenitor expansion, differentiation, cell-type specification, migration and maturation, early developmental insults will compound over time, leading to a circuit dysfunction.
Neurodevelopmental disorders (NDDs) pose such an example of disorders where early developmental insults from conception on result in a wide and heterogeneous spectrum of clinical diagnosis’s including intellectual disability (ID), autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD), schizophrenia (SCZ) and mood disorders (bipolar disorder (BD), major depressive disorder (MDD) . These NDDs are often diagnosed during childhood, and overlap between diagnostic categories . NDDs can be caused by both genetic and non-genetic sources. The most frequent non-genetic cause of NDDs is foetal alcohol syndrome disorder [3,4,5]. Additionally, the extreme genetic heterogeneity in NDDs is one of the major limiting factors in both diagnosis and treatment . With the use of sophisticated diagnostic tools such as whole exome sequencing and whole genome sequencing, the number of genes and variants linked to the aetiology of NDDs is vastly increasing [7, 8]. By doing so, chromatin remodelling genes have been found enriched in large datasets of NDD patients, and thereby pointing towards epigenetics as a convergent pathogenic mechanism [8,9,10,11,12,13]. By altering the epigenetic state of genes or histones, chromatin remodelers play an integral part in the machinery that translate external signals into lasting changes in gene expression patterns . Furthermore, chromatin remodelers are multifunctional proteins that can influence various processes across the developmental continuum, including neural progenitor generation and specification, cell-type differentiation and expansion, migration and circuit integration . Hence, as a significant subset of NDDs are caused by a failure of chromatin remodelling, it is to be expected that deficient chromatin remodelling will have a compounding effect across the developmental continuum, ultimately causing circuit dysfunction in mature networks .
Several reviews have already focussed on the most recent findings regarding the epigenetic origin of NDDs [17,18,19], however comparatively little is known about the function of these chromatin remodelers during the different programs of progenitor expansion, cell-type specification, neuronal migration and circuit integration. Using examples from mouse models mimicking these NDDs by introducing mutations in chromatin remodelers (also called NDD-related chromatinopathies), this review will first discuss how altered chromatin remodelling affects the different processes of the ongoing developmental continuum from mouse embryonic day (E) 10–17. Here, we will specifically focus on the three most abundant cell types found in the neocortex: excitatory (glutamatergic) and inhibitory (GABAergic) neurons, as well as glia cells. Although various chromatin remodelers have been described to play a role at one of these developmental processes, we chose to elaborate only on specific well-studied examples that play a role at multiple of these developmental steps, stressing their importance in neurodevelopment (Table 1). Furthermore, we will only focus on chromatin remodeler proteins and protein complexes and thereby exclude chromatin remodelling by non-coding RNAs (for a good review the authors would like to refer the readers to [20,21,22]).
Chromatin was first described by Walther Flemming for the unique fibrous structures observed in cellular nuclei . Chromatin is a highly dynamic structure that regulates the complex organization of the genome and thereby controls the gene expression underneath, and is composed of nucleosomes containing an octamer of histones (i.e. H2A, H2B, H3 and H4), wrapped by 147 base pairs of DNA and the linker histone H1 . The distinction between condensed heterochromatin and open euchromatin structures were first reported by Emil Heitz , and can be altered by chromatin remodelers via three distinct mechanisms, including: (i) sliding of an octamer across the DNA (nucleosome sliding), (ii) changing the conformation of nucleosomal DNA, and (iii) altering the composition of the octamers (histone variant exchange). By doing so, chromatin remodelling facilitates downstream gene expression in a cell-type and cellular demand-specific way, stressing their important role during (neuro) development.
Based on their function, three categories of chromatin remodelers have been classified, which are (i) the enzymes that control histone post-translational modifications (PTM) , (ii) DNA modifications that can attract/repel chromatin remodelling proteins or complexes, or (iii) enzymes that alter histone-DNA contact within the nucleosome via ATP hydrolysis . Furthermore, 3D genome architecture is increasingly considered an important epigenetic regulator of gene expression . In the next section, we will briefly discuss the global function of these epigenetic regulators (Fig. 1).
Histone modifying enzymes
Over the last decade a major effort was put into the identification of enzymes that directly modify histones. So far, enzymes have been identified for methylation , acetylation , phosphorylation , ubiquitination , sumoylation , biotinylation , ADP-ribosylation , deamination [36, 37], proline isomerization , β-N-glycosylation , crotonylation , propionylation , butyrylation , serotonylation , dopaminylation , Glutarylation [44,45,46], Lactylation , Benzoylation , S-palmitoylation , O-palmitoylation  and 5-Hydroxylysine . Also nonenzymatic histone PTMs have been identified including Glycation [52, 53], 4-Oxononanoylation [54, 55], Acrolein adduct [56, 57], Homocysteinylation [58, 59] nitrosylation [60,61,62,63], sulfe‐, sulfi‐, and sulfonylation [64, 65] and S-glutathionylation [66, 67]. For many of them, also enzymes have been identified that can remove the PTM, or ‘read’ the PTM and recruit other proteins to form a chromatin remodelling complex (for a detailed review of all histone modifying enzymes and their function see [26, 68]). Currently, two mechanisms are known by which histone PTMs can alter the state of chromatin. First, it is accepted that all histone modifications have the potential to affect higher order chromatin structure by neutralising the basic charge of the nucleosome, and therefore could loosen inter or intra-nucleosomal DNA-histone interactions [69,70,71,72]. A well-known example is acetylation of lysine residues, which removes the positive charge of lysine and therefore increases the probability to alter the structural state of chromatin . Second, histone PTMs can recruit non-histone proteins to set in motion processes such as transcription, DNA repair and DNA replication . Depending on which histone modification (or which sequence of histone modifications) are present at a given histone tail, different sets of proteins are encouraged to bind or prevent from binding to chromatin. This recruitment process is highly dynamic, as multi-step processes (e.g. transcription) require a distinct set of histone PTMs to recruit a distinct set of chromatin-remodelling proteins. Proteins that are recruited to PTMs contain specific PTM-reader domains (such as bromo-domains, chromo-(like) domains, PhD domains etc.). Many chromatin remodelers actually possess multiple reader domains, suggesting their ability for multivalent interactions that would increase both affinity and specificity [74, 75]. Functional interplay among writer-eraser PTM enzymes in the brain remains largely unknown. Recent reports, however, showed that knockout mice of the writer-eraser duo Kmt2a and Kdm5c, which are responsible for Wiedemann-Steiner Syndrome and Mental Retardation X-linked Syndromic Claes-Jensen, share similar brain transcriptomes, cellular- and behavioural deficits . Double mutation of Kmt2a and Kdm5c however partly corrected H3K4 transcriptomes as well as their cellular and behavioural deficits, suggesting this balance is essential during development and might be an interesting therapeutic strategy for NDDs .
In addition to methylation of histone tails, DNA can also be methylated to regulate chromatin state transitions (Fig. 1). The addition of a methyl group from S-adenosyl-L-methionine substrates only occurs on cytosines that are followed by guanines (called CpG sites), and is catalysed by DNA methyltransferases (DNMTs) leading to gene repression. There are two types of DNMT classes, namely either the de novo methyltransferases or maintenance methyltransferases . DNMT3a and DNMT3b are classified as de novo methyltransferases as these can methylate previously unmethylated cytosine of CpG dinucleotides on both strands . DNMT1 is classified as a maintenance methyltransferase as it has a substrate preference for hemimethylated DNA over unmethylated DNA. In contrast to its preference, DNMT1 can also display de novo methyltransferase activity in a specific cellular context-dependent manner . DNA methylation can affect chromatin remodelling either by attracting transcriptional activators to the methylated cytosine , or it can attract transcriptional repressors that have methyl cytosine-binding domains. For example, DNA methylation can recruit histone deacetylases, which facilitate the formation of the silent chromatin state . These methyl cytosine-binding proteins include methyl CpG-binding domains (MBDs) [82, 83] and methyl CpG-binding protein 2 (MeCP2) , which are known for their role in the aetiology of NDDs. During human development, genomic DNA methylation signatures are established in early development by two consecutive waves of nearly global demethylation, followed by targeted re-methylation [85, 86]. NDD mutations have often been found to underlie errors in methylation during these early time points [87, 88]. Consequently, altered methylation signatures during early development may be passed on across all cell lineages and can thus affect multiple tissues. When these epigenetic changes are maintained throughout development and across cell-types, these so called ‘episignatures’ can be used as biomarkers for the diagnosis of NDDs using easily accessible tissues such as peripheral blood [89,90,91,92]. Indeed, several very recent studies have already showed the potential for using disease-specific episignatures as diagnostic tool for NDDs, including patients with a known diagnosis as well as patients carrying variants of unknown significance [93,94,95].
ATP-dependent chromatin remodelers
The third class of chromatin modifying enzymes are the ATP-dependent chromatin remodelers including SWI/SNF (switch/sucrose-non-fermenting), ISWI (imitation switch), CHD (chromodomain-helicase-DNA binding) and INO80 (inositol requiring 80). In general, ATP-dependent chromatin remodelers hydrolyse ATP to generate enough energy to disrupt the interactions between histones and DNA. By doing so, ISWI remodelers can alter nucleosome positioning to promote heterochromatin formation and thus transcriptional repression. The SNF2/SNF chromatin remodelling family act as DNA translocases, by destroying histone-DNA bonds forming a DNA loop that propagates around the nucleosome until it reaches the exit site on the other side of the nucleosome . Furthermore INO80 chromatin remodelers in vivo have been shown to play a role in nucleosome eviction, and histone variant exchange of the histone dimer H2A-H2B by the H2AZ-H2B dimer . Finally, CHD family members exert a heterogeneous set of biological properties. One of the best studied examples is the NURD (nucleosome remodelling and deacetylase) complex, which contains lysine‐specific histone demethylase 1A (LSD1), Chromodomain Helicase DNA Binding Protein 3 (CHD3) or CHD4, histone deacetylases (HDAC1 or HDAC2) and methyl CpG-binding domain (MBD) proteins. The NURD complex has been shown to deacetylate specific gene sets during development leading to transcriptional repression .
ATP-dependent chromatin remodelers all share a conserved core ATPase domain, however all ATP-dependent chromatin remodelers harbour exclusive domains adjacent to the ATPase domain . Each of these domains play a role in the recruitment of remodelers to chromatin, interaction with specific histone modifications and/or are involved in the regulation of the ATPase activity of the remodelers (see  for a detailed review on their function).
3D chromatin architecture
3D genome architecture is increasingly considered as an important epigenetic regulator of gene expression. On a coarse level, genomes are organised into structures known as chromosome territories (Fig. 1) . These chromosome territories separate euchromatic from heterochromatic regions, and are termed ‘A’ and ‘B’ compartments, respectively . Within the chromosome territories, megabase-sized chromatin domains are organised into smaller and smaller sub-domains known as topologically associated domains (TADs) . TADs can be found in either ‘A’ or ‘B’ compartments, and are separated by sharp boundaries across which contacts are relatively infrequent. Interestingly, the boundaries between TADs are strikingly consistent across cell divisions and between cell types, as roughly 50%–90% of TAD boundaries have been shown to overlap in a pairwise comparisons between cell types . In ‘A’ compartments, TADs are regions where DNA is highly organised in 3D space to enable “long-range” transcriptional regulation. This long-range transcriptional regulation is possible because enhancers are in close physical proximity to the promoters of their target genes in 3D space, despite long stretches of intervening nucleotides . This physical proximity allows protein complexes bound at enhancers to interact with those bound at promoters (i.e. called enhancer-promotor loops), thereby influencing transcription of target genes. CCCTC-binding factor (CTCF) and Cohesin are such proteins that facilitate chromatin looping interactions . CTCF-mediated loop formation requires one CTCF at each end of the chromatin loop, which dimerize if they are facing each other in the opposite orientation . CTCF interacts with Cohesin via its C-terminal tail  and may thus allow Cohesin to locate on a particular side of the interaction to anchor and stabilize the chromatin loop . In addition to Cohesin and CTCF, other proteins such as YY1 , ZNF143 and Polycomb group proteins , repetitive elements and PTMs are enriched at TAD boundaries to support transcription. These repetitive elements at TAD boundaries have been found to act as specific anchor points to spatially organize chromosomes , whereas enrichment for the transcription marks H3K4me3 and H3K36me3 in TAD boundaries show an association with highly expressed regions and suggests that transcription itself plays a role in TAD organisation . By doing so, 3D chromatin structures play an essential role in orchestrating the lineage-specific gene expression programs that underlie cellular identity .
In summary, the above examples show that cells possess a wide range of chromatin remodelers to translate external signalling cues into lasting changes in gene expression (Fig. 1). Interestingly, chromatin remodelers are especially strongly expressed in the brain , and it is therefore not surprising that impaired chromatin remodelling in any of the above described remodeler classes has been identified to a cause monogenic forms of NDDs [68, 112,113,114]. Chromatin remodelers are often multifunctional , and by doing so have the ability to play divergent roles in the multi-step continuum of brain development. This continuum encompasses neural progenitor generation and specification, cell-type differentiation and expansion, migration and circuit integration. Dysfunction of chromatin remodelers at any point during this developmental continuum will result in lasting changes on mature network function. In the next section we will discuss the different steps along this developmental continuum, and explain how altered chromatin remodelling at any of these steps ultimately affects the structure and function of mature neuronal networks.
Epigenetic modulation during neurodevelopment and disease
In the early stages of neocortical development, the telencephalic wall is composed neuroepithelial (NE) cells that will give rise to diverse pools of progenitor cells . As these progenitors proliferate and expand in number, some begin to differentiate into radial glia cells (RGCs), establishing the ventricular zone (VZ). RGCs in turn begin to produce projection neurons and intermediate progenitors (IP) around E11.5 in mice which establish the subventricular zone (SVZ) . In human development, RGCs not only produce projection neurons and IPs, but also the human specific outer radial glia (oRG) between gestational week (GW) 16–18, which will populate the outer SVZ (oSVZ) . The RGCs in mice and oRG in humans act as transit-amplifying cells to increase the population of glutamatergic neurons until E18 in mice  or GW 21 in the human neocortex [119, 120], after which they switch to local glia production  (Fig. 2).
Epigenetic modulation during neocortical development
Mutations in the Polycomb repressive complex (PRC) are a prime example of how defective chromatin remodelling influences progenitor proliferation. The PRC consists of two complexes: PRC1 and PRC2. Although for PRC1 no dominant germline mutations have been described, autosomal recessive mutations in the PRC1 complex protein PHC1 have been shown to cause a form of microcephaly with short stature in two Saudi siblings . Loss of PHC1 resulted of the inability of patient cells to ubiquitinate H2A, resulting in increased Geminin expression causing cell cycle abnormalities and impaired DNA damage response . Additionally, mutations of interactors of PRC1 are mutually vulnerable to cause NDDs with brain volume abnormalities. For example, mutations in the PRC1 interactor AUTS2 have been shown to cause an autosomal dominant form of syndromic mental retardation, including comorbidities such as ID, ASD, microcephaly, short stature and cerebral palsy [123, 124]. Mouse embryonic stem cells (mESCs) carrying heterozygous mutations in Auts2 showed an increase in cell death during in vitro corticogenesis, which was rescued by overexpressing the human AUTS2 transcripts. Furthermore, mESCs harbouring a truncated AUTS2 protein (missing exons 12–20) showed premature neuronal differentiation, whereas cells overexpressing AUTS2 showed increase in expression of pluripotency markers and delayed differentiation.
The PRC2 complex consists of core subunits Enhancer of Zeste 2 (EZH2) and its homolog EZH1, which catalyse mono‐, di‐, and tri‐methylation of H3K27 resulting in heterochromatin formation and gene repression. EZH2 is mainly expressed during embryogenesis, while EZH1 is more ubiquitously expressed in adult and quiescent cells . Loss-of-function mutations in EZH2 and thus reduced H3K27me3 levels in humans have been shown to cause Weaver syndrome, causing overgrowth and macrocephaly, accelerated bone maturation, ASD, developmental delay and characteristic facial features [126,127,128]. In accordance with the overgrowth phenotype in Weaver syndrome, conditional KO of Ezh2 in mice accelerated proliferation of neuronal precursors in the cerebral cortex at the expense of self-renewal of progenitors . These results were strengthened by showing that fate-mapped E14-born neurons in a cKO for the PRC2 complex member Eed (which interacts with EZH2) mainly resided in the layer 2/3, in contrast to WT E14- born neurons, which resided in layer 4 . These results confirmed PRC2 regulates the progression of apical progenitor’s temporal specification [130, 131].
Progenitors follow a specific pattern of cell divisions to initiate the build-up of the different layers in the cortex . This is done in an ‘inside-out’ fashion, meaning the early-born neurons form the deep neocortical layers (i.e. 6 and 5), whereas late born neurons radially migrate through the deeper layers to create the more superficial layers (layer 4 and 2/3). Each RGC has been shown to consistently produce 8 to 9 glutamatergic neurons progressing from lower-layer excitatory neurons at mouse E12.5 to glutamatergic neurons that undergo radial migration towards the upper-layers at mouse embryonal day E15.5 . Approximately 1 in 6 RGCs switches to gliogenesis, generating astrocytes and oligodendrocytes at the end of the cell cycle . The process of neuroprogenitor specification during corticogenesis has been studied and reviewed in great detail elsewhere (see  or for a review see ) and will therefore not be discussed in this review. Numerous studies however have shown that defects in chromatin remodelers during RGC specification result in shifts in the neuron classes produced or in precocious cell cycle exit and gliogenesis, and thereby could contribute to the neuronal phenotypes found in NDD patients . In mice, Ezh2 is highly expressed in RGCs up to E14.5 and has been proposed to regulate RGC identity and proliferation behaviour, as well as RGC‐to‐glial‐progenitor transition [129, 136, 137] by inhibiting Neurogenin 1 expression . Ablation of Ezh2 in mouse RGCs correlates with premature RGC differentiation as described earlier, increased generation of lower‐layer neurons, decreased upper‐layer neuron production, and precocious astrocyte generation . Furthermore, knockdown of Ezh2 in mice has been shown to affect the neuronal polarization and radial neuronal migration .
In addition to PRC2, deletion of the PRC1 component Ring1b at E13 prolongs the expression of Fez transcription factor family member zinc-finger 2 (Fezf2). The prolonged expression of Fezf2 results in the continuous expression of downstream target genes such as Ctip2, resulting in an increased production of deep-layer neurons . Interestingly, when Ring1b is knocked out later in developmental time at E14.5, the number of upper-layer neurons is increased instead, due to an extended neurogenic period .
Similar to H3K27me, H3K9me2/3 is a repressive mark which is established by the methyltransferases SETDB1, (KMT1E), SUV39H1 (KMT1A), G9a (EHMT2, KMT1C) and G9a‐like protein (EHMT1, KMT1D). SETDB1 has been associated to several NDDs. In humans, two missense mutations  and a microdeletion  in SETDB1 were found in a large cohort of ASD patients. In addition, SETDB1 has been implicated to play a role in the aetiology of Schizophrenia by regulating GRIN2B expression . Moreover, SETDB1 has been shown to influence chromatin 3D structure by binding to a non-coding element upstream of the Pcdh cluster [143, 144] which was a near-perfect match to a Schizophrenia risk haplotype (number 108 in ). Finally, SETDB1 is also described to play an indirect role in Prader-Willi syndrome, by contributing to the maternal silencing of the SNORD116 gene [145, 146]. Setdb1 is highly expressed early in corticogenesis (E9.5) in NE cells in the VZ, however its expression declines at E15.5 . While deletion of Setdb1 does not affect RGC numbers, it has been shown to reduce layer V and VI Ctip2+ basal progenitors between E14.5 and E16.5 . At the same time, an increase of the number of Brn2+ layer II and III neurons was found in the CP. This shift in the production of upper layer neurons at the expense of deep layer neurons remained after neurogenesis ceased at E18.5. All these events together lead to a reduced cortical volume in Setdb1 knockout mice at E18.5 and P7 . Moreover, deletion of Setdb1 causes accelerated astrogliogenesis, demonstrating that Setdb1 does not only regulate the timing of late neurogenic events, but also the RGC‐to‐astrogenic‐progenitor transition . As SETDB1 is a H3K9 methyltransferase, its general role is to repress gene transcription, or methylate DNA. SETDB1 does so by being involved in several complexes. SETDB1 interacts with The Krüppel-associated box (KRAB) domain-containing zinc-finger proteins (KRAB-Zfp) and KRAB domain-associated protein 1 (KAP-1) , which recruit the NuRD complex and HP1 to form a repressive complexes . SETDB1 can also interact with MBD1 and ATF7IP , which has been suggested to play a role in X-inactivation . To repress gene transcription via DNA methylation, SETDB1 interacts with DNMT3A/B in cancer cells, to represses the expression of p53BP2 and RASSF1A . Through the interaction with these complexes, SETDB1 can thus target various genomic loci in different cell types, at different stages during brain development.
Like SETDB1, EHMT1 plays a major role in embryonic development, as full knockout of this gene leads to embryonic lethality . In hematopoietic stem cells, EHMT1 activity is essential to facilitate the long-term silencing of pluripotency genes, and thus inhibition of EHMT1 is essential for maintaining pluripotency . Furthermore, a reduction of large H3K9me2 chromosome territories was found in these stem cells , which are proposed to stimulate lineage specification by affecting the higher order chromatin structure . Little is known about the role of EHMT1 during early neurodevelopment, however knockdown of EHMT1 in NPCs revealed differential expression of genes important in development, such as BMP7, WNT7A, CTNNB1, TGFB2 and CHD3 . Furthermore, in neural progenitors EHMT1 has been found to interact with ZNF644 to silence multipotency and proliferation genes. Disruption of the ZNF644/EHMT1 resulted in maintenance of proliferative identity and delayed formation of differentiated retinal neurons . Similar roles for EHMT1 in the regulation neural progenitor genes were found in a conditional knockout mice (Camk2a-Cre; GLPfl/fl) . Interestingly, no brain volume abnormalities are described in animal models of EHMT1 haploinsufficiency [158, 159] whereas microcephaly was found in 20% of patients carrying intragenic EHMT1 mutations . These patients are diagnosed with Kleefstra Syndrome, which is in addition to microcephaly characterised by mild to severe ID, ASD, developmental delay, speech problems, hypotonia, characteristic facial features, epileptic seizures, heart defects and various behavioural difficulties [160, 161].
Histone acetylation by HATs and removal by HDACs is another example where deficits in chromatin remodelling affect the cellular distribution during corticogenesis. For example, the gene encoding the HAT cAMP‐response element binding protein (CBP) is highly expressed in proliferating RGCs and post‐mitotic neurons during corticogenesis , and CBP null mice have been shown embryonically lethal due to failure of neural tube closure (E9-E12.5) , stressing their role in neurodevelopment . In humans, heterozygous mutations in CBP are associated with Rubinstein–Taybi syndrome (OMIM# 180849), which is characterized by ID, postnatal growth deficiency, microcephaly, broad thumbs and halluces, and dysmorphic facial features . In mice, similar brain volume abnormalities have been described . CBP induces acetylation of H3K9, H3K14 and H3K27 within target gene promoters, such as α1‐tubulin (E13‐E16), Gfap (E16‐P3), S100β, Plp2 and Mbp (P14) . Accordingly, heterozygous loss of Cbp in mice diminishes acetylation at these promoters and leads to decreased differentiation of progenitor towards astrocytes and neurons . Consequently, a reduced number of neurons, astrocytes and oligodendrocytes were observed in the cortex of heterozygous Cbp mice around P14, whereas an increase in PAX6 expressing progenitors was found as compared to wild-type mice . At later ages (P50), only a reduction in gliogenesis remained in the corpus callosum . Another example of a HAT important in NPC development is Lysine acetyltransferase 8 (KAT8), a member of the Non-specific Lethal (NSL) complex which is responsible for acetylation of H4K16 and plays a role in H4K16 propionylation . Cerebrum specific knockout of Kat8 in mice has recently been shown to cause severe cerebral hypoplasia in the neocortex and in the hippocampus, together with postnatal growth retardation and pre-weaning lethality . Furthermore, these mice showed a loss of RGC proliferation and thus reduced progenitor pool at E13.5, massive apoptosis starting at E12.5 and increased numbers of Tuj1+ cells, indicating precocious neurogenesis at E13.5 . Similarly, patients with KAT8 mutations presented with variable brain MRI abnormalities, epilepsy, global developmental delay, ID, facial dysmorphisms, variable language delay, and other developmental anomalies . Interestingly, patients with epilepsy responded well to the histone deacetylase inhibitor Valproate , stressing the importance of KAT8 and other lysine acetyltransferases function in brain development.
Within the NSL complex, KAT8 is regulated by the NSL regulatory subunits KANSL1 and KANSL2. Mutations in KANSL1 cause Koolen-de Vries Syndrome (OMIM# 610443), characterized by ID, distinctive facial features, and friendly demeanour . Likewise, KANSL2 mutations have been identified in ID patients . Interestingly, haploinsufficiency of Kansl1 in the mouse causes craniofacial abnormalities, reduced activity levels and impaired fear learning, as well as epigenetic dysregulation in genes linked to glutamatergic and GABAergic cells .
The balance between acetylation and deacetylation plays an important role in progenitor proliferation and differentiation, as inhibition of HDAC activity also results in progenitor proliferation/differentiation deficits . Histone acetylation is removed by HDACs, resulting in chromatin condensation and transcriptional repression . So far, over 18 HDACs are characterised in the mammalian genome, and they are expressed in a cell type- and developmental stage-dependent fashion . For example, HDAC1 is highly expressed in neural stem cells/progenitors and glia, whereas HDAC2 expression is initiated in neural progenitors and is up-regulated in post-mitotic neuroblasts and neurons, but not in fully differentiated glia . Conditional knockout of Hdac1 or Hdac2 in mice progenitors impairs neuronal differentiation . Specifically, conditional knockout of Hdac1 and Hdac2 in Gfap-Cre mice resulted in major brain abnormalities and lethality at around P7 , whereas conditional knockout of Hdac1 and Hdac2 in Nestin-Cre mice resulted in reduced proliferation and premature differentiation of NPCs prior to abnormal cell death . Moreover, inhibition of HDAC activity at the neurogenic stage decreases the production of deep-layer neurons and increases the production of superficial-layer neurons . Conditional deletion of Hdac1 and Hdac2 in oligodendrocyte precursors results in severe defects in oligodendrocyte production and maturation . Recently, the first patient carrying a mutation in HDAC2 has been characterised presenting with many clinical features consistent with Cornelia de Lange Syndrome (CdLS) including severe developmental delay, limb abnormalities, congenital heart defects, altered development of the reproductive system, growth retardation and characteristic craniofacial features . No patients harbouring mutations in HDAC1 have been characterized as to date.
Next to HDAC2, missense mutations in HDAC8 are also linked to the aetiology of CdLS presenting with overlapping clinical features (OMIM# 300269) [178, 179]. HDAC8 plays a key role in regulating cohesion function by deacetylating one of the core cohesion proteins, SMC3, which affects mitosis as well as transcription through loss of TAD function [180, 181]. Loss of HDAC8 activity in SVZ progenitors from 4-month old mice has been shown to reduce progenitor proliferation and differentiation . Moreover, knockdown of HDAC8 in the mice embryonic carcinoma cell line P19 cells permitted the formation of embryoid bodies . Furthermore, loss of HDAC8 in zebrafish has been found to increase apoptosis in CNS progenitors . Recently a novel intronic variant in HDAC8 was found in a large Dutch family with seven affected males presenting with X-linked ID, hypogonadism, gynaecomastia, truncal obesity, short stature and recognisable craniofacial manifestations resembling but not identical to Wilson-Turner syndrome (OMIM# 309585) . This variant disturbs the normal splicing of exon 2 resulting in exon skipping, and introduces a premature stop at the beginning of the HDAC catalytic domain . How this specific variant influences neurodevelopment remains elusive.
Chromatin 3D organisation
The 3D organization of chromatin is changing dynamically as the cell differentiates. Whereas the nuclei of embryonic stem cells have been shown to be relatively homogenous, heterochromatin foci are becoming more apparent during differentiation into progenitors. When these progenitors differentiate into neurons, the heterochromatin foci are becoming even larger, suggesting that heterochromatin regions are actively reorganized during differentiation . Deficiencies of 3D chromatin organizers such as CTCF and Cohesin are associated with the aetiology of NDDs called CTCF-associated NDDs and cohesinopathies, respectively. Heterozygous mutations in CTCF (OMIM# 604167) have been shown to cause the NDD mental retardation, autosomal dominant 21 (MRD21), which is characterised by variable levels of ID, microcephaly, and growth retardation [186,187,188]. In mice, loss-of-function studies of Ctcf revealed an important role for this protein in cell fate specification and neural differentiation. Knockout of Ctcf at E8.5 resulted in upregulation of PUMA (p53 upregulated modulator of apoptosis), leading to high levels of apoptosis and loss of the telencephalic structure . Inactivation of CTCF several days later (E11) also resulted in PUMA upregulation and increased apoptotic cell death, and again the CTCF-null forebrain was hypocellular and disorganized at birth . In contrast, conditional knockout of CTCF in postmitotic projection neurons resulted in misexpression of clustered protocadherin (Pcdh) genes leading to altered functional neuronal development and neuronal diversity . These results suggest that CTCF activity regulates the survival of neuroprogenitor cells, and the balance between neuroprogenitor cell proliferation and differentiation .
The two best-described cohesinopathies are CdLS (OMIM# 122470, 300590, 610759, 614701, 300882) and Roberts syndrome (and its variant SC Phocomelia, OMIM# 268300). As described above, CdLS can be caused by loss of function mutations in HDAC2 and HDAC8. Additionally, mutations in three Cohesin subunits (SMC1α, SMC3, RAD21) [191, 192] and in one Cohesin-interacting protein (NIPBL)  have been found causal for CdLS, whereas mutations in ESCO2 are responsible for Roberts syndrome/SC Phocomelia (OMIM# 269000) . While a full knockout of Cohesin subunits in mice is lethal, mice carrying heterozygous mutations in these genes are viable and show altered gene expression in developmental programs, DNA repair and replication .
Chromatin remodelling complexes: CHD proteins and the NuRD complex
Altered chromatin remodelling by multi-subunit protein complexes has been shown to play a role in RGC differentiation and in the aetiology of NDDs. One of these complexes, called the NuRD complex, consists amongst other proteins of LSD1, HDAC1/2, and a Chromodomain Helicase DNA (CHD) Binding Protein (either CHD3, 4 or 5) . These CHD proteins have been shown to play essential roles during neurodevelopment, as pathogenic variants in CHD1, CHD2, CHD3, CHD4, CHD6, CHD7 and CHD8 have been associated with a range of neurological phenotypes. Of the nine human CHD family members that have been characterized (CHD1-9), further subdivisions are being made into subgroups based on their function. Subfamily one consists of CHD1 and CHD2 because of their shared DNA binding domain that is not well-conserved in the other CHD proteins . CHD1 has been found to play an essential role in early mice development, as Chd1−/− embryos show proliferation defects and increased apoptosis, are smaller than controls by E5.5 and fail to become patterned or to gastrulate . Similar results in decreased self-renewal and pluripotency were found using knockdown of Chd1 in mESC cells . Furthermore, Chd1−/− ESCs show deficits in self-renewal and a reduction in genome-wide transcriptional output by directly affecting ribosomal RNA synthesis and ribosomal assembly . In contrast, mice lacking a single Chd1 allele (Chd1+/−) are healthy, fertile and phenotypically normal .
Recently the first patients with CHD1 missense mutations were identified, which presented with ID, ASD, developmental delay, speech apraxia, seizures, and dysmorphic features. Interestingly, also a patient with a microdeletion spanning RGMB and the last exons of CHD1 was characterised with no obvious NDD phenotype, suggesting that whereas deletions of CHD1 may not cause a consistent neurological phenotype, missense mutations in CHD1 may do so via a dominant negative mechanism .
Despite the fact that CHD family members are rather ubiquitously expressed, only CHD2 pathogenic variants cause a brain-restricted phenotype, suggesting a unique role for this gene in neurodevelopment. Based on loss-of-function studies, CHD2 has been shown to regulate self‐amplification of RGCs and prevents precocious cell-cycle exit. CHD2 is mainly expressed in RGCs between E12-E18 and rarely in IPs, however knockdown of Chd2 in utero resulted in a reduction of RGCs in the SVZ whereas an increase was found in the number of produced IPs and neurons (Fig. 2) [201, 202]. This premature differentiation in RGCs can lead to a depletion of the progenitor pool, resulting in a smaller overall brain volume as a consequence . Indeed patients harbouring mutations in CHD2 present with a reduced head size and in 20% of the cases microcephaly [204, 205], developmental delay, ID, ASD, epilepsy and behavioural problems with phenotypic variability between individuals . A subset of patients carrying CHD2 pathogenic variants present with developmental and epileptic encephalopathy (also called Dravet Syndrome), which is an early onset of epilepsy disorders characterized by refractory seizures and cognitive decline or regression associated with ongoing seizure activity .
CHD3, CHD4 and CHD5 are categorised in the second CHD family because they share dual plant homeodomain zinc finger domains . Furthermore, these class 2 CHD remodelers exhibit subunit‐specific functions and display mutually exclusive occupancy within the NuRD complex at different stages of corticogenesis . CHD3, has been shown to play an important role in the correct cortical layering and controls the timing of upper‐layer neuron specification . In mice, CHD3 expression starts around E12.5 where it is still low expressed, and this increases from E15.5 to E18.5. At these later time points, CHD3 is mainly expressed in upper layer neurons in the cortical plate. Neurons lacking CHD3 (CHD3-knockdown) have been found more likely to express transcription factors that regulate laminar positioning and differentiation of deeper cortical layers (i.e. Tbr1 and Sox5), whereas a lower number of neurons expressed the upper layer markers Brn2 and Cux1, implicating that CHD3 may influence the expression of genes that couple radial migration with laminar identity (Fig. 2) . Patients with CHD3 mutations are only very recently identified with Snijders Blok–Campeau syndrome, which is characterized by ID (with a wide range of severity), developmental delays, macrocephaly, impaired speech and language skills, and characteristic facial features [209, 210].
The second CHD protein that plays a role in the NuRD complex is CHD4. Mice lacking Chd4 almost always died at birth, however when examining brain volumes at E18.5 a significant reduction of the cortical thickness was found, caused by reduced NPC proliferation and premature cell cycle exit, followed by increased apoptosis of premature born neurons . As a result, CHD4fl/fl/Nestin-Cre mice presented with lower numbers of IPs and late born upper layer neurons (Fig. 2) . Interestingly, Chd4 appears to guide Polycomb repressor complex (PRC2) and especially Ezh2 to opposing effects early vs. late in corticogenesis, first interacting to repress the gliogenic gene Gfap, and later repressing the neurogenic Ngn1 after the neurogenic-to-gliogenic switch . Interestingly, patients carrying mutations in CHD4 actually present with macrocephaly, amongst other characteristics like ID, hearing loss, ventriculomegaly, hypogonadism, palatal abnormalities and facial dysmorphisms that are diagnosed by Sifrim–Hitz–Weiss syndrome . The opposing phenotypes found for brain volume between rodent models and patients might be explained by a gene dosage effect, as for some variants in CHD4 altered ATPase activity levels were found, suggesting a possible gain-of-function phenotype in certain patients [212, 213]. Another possibility might be that the NuRD complex function is differentially regulated in humans and mice, as was recently suggested in a study comparing mouse and human pluripotent stem cells .
Finally, CHD5 has only recently been characterized as one of the core subunits of the NuRD complex . CHD5 is the only CHD member that is mainly expressed in the total brain, foetal brain and cerebellum . Chd5 expression was mainly found in late-stage neuronal progenitors undergoing terminal differentiation, rather than in proliferating progenitors . In utero knockdown of Chd5 furthermore resulted in an accumulation of undifferentiated progenitors, which were unable to exit the VZ, SVZ and intermediate zone (IZ; Fig. 2) . Additionally, knockdown of Chd5 in mouse ESCs resulted in a failure to upregulate late stage neuronal genes . Similar results for the role of Chd5 in neuronal gene regulation were found in primary rat cultures . A de novo damaging missense variant in the CHD5 gene was identified in an ASD proband from the Autism Sequencing Consortium . In accordance, knockout of Chd5 in mice indeed has been shown to cause ASD-like behaviour including increased anxiety and decreased social interaction .
The third subfamily consists of the remaining family members CHD6-9 . CHD6 mutations have previously been described, including a large translocation in one Pitt-Hopkins patient , in a single case of mental retardation , in sporadic acute myeloid leukaemia incidences , and most recently for the very rare Hallermann–Streiff syndrome . CHD6 is the least studied member of the CHD family, and little is known for its contribution during neurodevelopment.
Similar as to other members of the CHD family, loss of CHD7 resulted in impaired proliferation and self-renewal of RGCs in the SVZ. Consequently, a reduction of NE thickness in telencephalon and midbrain was shown in Chd7 homozygous gene-trap mutant embryo at E10.5 . Similarly, mESCs from Chd7Gt/Gt mice showed a reduced potential to differentiate into neuronal and glial lineages, and presented with altered accessibility and expression of NPC genes. Furthermore, neurons generated from these Chd7Gt/Gt mESCs presented with a significant lower length and complexity . Furthermore, CHD7 plays a key role in oligodendrocyte precursor survival and differentiation (Fig. 2) [226, 227] and has been shown to cooperate with Sox10 to regulate myelination and re-myelination . Additionally, CHD7 has been shown to play an important role in cerebral granular cell differentiation and cell survival .
Mice lacking one Chd7 copy (found in Chd7COA1/+ mice  and Chd7Gt/+ mice ) also often present with brain abnormalities including the absence or hypoplasia of olfactory bulb, cerebral hypoplasia, defects in the development of telencephalic midline and reduction of the cortical thickness [229, 230]. Patients with CHD7 mutations present with similar brain structure abnormalities including hypoplasia of olfactory bulb and cerebellum, agenesis of the corpus callosum, microcephaly and atrophy of the cerebral cortex [231,232,233]. Additionally, in relation to its role in oligodendrocyte differentiation and function, some patients have been characterised with white matter defects [234, 235]. Loss of CHD7 in patients is called CHARGE syndrome (OMIM# 214800), which is next to the brain malformations characterised by coloboma, heart defects, growth retardation, genital hypoplasia, and nose and ear abnormalities (including choanal atresia, deafness and vestibular disorders) .
Finally, CHD8 has been identified as a causal gene for ASD, presenting with common phenotypic features included macrocephaly, accompanied by rapid early postnatal growth, characteristic facial features, increased rates of gastrointestinal complaints and marked sleep dysfunction . Chd8 is strongly expressed around the transition from symmetric proliferative to asymmetric neurogenic RGC division , and knockdown of Chd8 at E13 has been shown to prematurely deplete the neural progenitor pool in developing mice cortices (Fig. 2) . Similarly, both heterozygous and homozygous knockout of Chd8 in mESCs resulted in an upregulation of neuronal genes upon differentiation into NPCs . Indeed, CHD8 binds the promoters of cell cycle genes and serves as a transcriptional activator of for example PRC2 components Ezh2 and Suppressor of Zeste 12 , which allows for the repression of neural genes during this developmental period . In human iPSC-derived CHD8+/− organoids, a number of ASD risk genes was upregulated . Furthermore, CHD8 is identified as a negative regulator of the Wnt–β-catenin signalling pathway , as knockdown of Chd8 in non-neuronal cells lead to an enrichment of up-regulated Wnt–β-catenin signalling pathway genes . Interestingly, knockdown of Chd8 in neuronal cells lead to an enrichment of down-regulated Wnt–β-catenin signalling pathway genes, indicating CHD8 plays cell-type specific roles . Furthermore, conditional knockout of CHD8 in oligodendrocytes (Chd8flox/flox;Olig1-Cre+/−) has shown to impair oligodendrocyte differentiation and myelination in a cell-autonomous manner [243, 244]. Whereas homozygous deletion of Chd8 in mice results in early embryonic lethality , Chd8 heterozygous mice were viable and presented with similar phenotypes as patients, including macrocephaly, abnormal craniofacial features, and ASD like behaviour [246,247,248,249,250]. Moreover, introducing the human truncating variant N237K into the mouse Chd8 gene (Chd8+/N2373K) revealed ASD-like behaviour, aberrant vocalization, enhanced mother attachment behaviour and enhanced isolation-induced self-grooming specifically in males, but not females . These phenotypes were also conserved in zebrafish, where Chd8 knockdown was found to cause macrocephaly and gastrointestinal phenotypes [237, 238].
Chromatin remodelling complexes: SWI/SNF
Also SWI/SNF chromatin remodelling complex subunits are expressed in a temporal and cell‐type specific manner [199, 252]. During differentiation from embryonic stem cells into neurons, the SWI/SNF complex begins to switch subunits to those unique to neural progenitors, followed by subunits specific to neurons [253, 254]. Neural progenitor proliferation requires a SWI/SNF complex containing PHF10 and ACTL6A subunit, which are replaced by the subunits DPF1, DPF3, and ACTL6B when neural progenitors exit the cell cycle to become post-mitotic neurons [255, 256]. Interestingly, the neural progenitor-specific SWI/SNF complex exclusively incorporates either SMARCC2 or SMARCC1 subunits at distinct developmental stages. In the mouse, neural progenitor SWI/SNF complexes harbours SMARCC2 until E14.5 to repress intermediate progenitor generation, whereas between E14.5 and E15.5, SMARCC2 is replaced by SMARCC1, activating intermediate progenitor generation in RGCs via the interaction with the H3K27 demethylases JMJD3 and UTX [257,258,259]. Double loss of Smarcc2 and Smarcc1 from as early as E10.5 (Smarcc1fl/fl:Smarcc2fl/fl, Emx-Cre) resulted in reduced numbers of proliferative progenitors, thinning of the cortical SVZ and loss of projection neurons . In addition, loss of Smarcc2 and Smarcc1 in late NPCs (Smarcc1fl/fl:Smarcc2fl/fl, hGFAP-Cre) resulted in H3K27me3-mediated silencing of neuronal differentiation genes, causing delayed cortical and hippocampal neurogenesis . Similarly, a loss of the catalytic subunit Smarca4 in late NPCs (Smarca4fl/fl::hGFAP-Cre) resulted in reduced cortical thickness, dendritic abnormalities, hippocampal underdevelopment and cerebellar disorganization . Furthermore, Smarcc2 and Smarcc1 double knockouts showed an upregulation of Wnt signalling activity resulting in increased progenitor proliferation-related defects . This Wnt /β-catenin pathway has indeed previously been shown to be a critical regulator of NPC proliferation and neurogenesis during cortical development [116, 262, 263]. Thus, timely expression of these SWI/SNF subunits  is essential for regulating cell fate during neurodevelopment, and can control this processes by regulating Wnt signalling activity .
Not only SMARCC1 and SMARCC2 expression is essential during brain development, also divergent patterns of expression of SMARCA1 and SMARCA5 were found in the mouse embryo. Whereas Smarca5 is mainly expressed in proliferating progenitors in the neocortex and the cerebellum, Smarca1 is predominantly expressed in terminally differentiated neurons after birth in the cerebellum and hippocampus of adult animals [265, 266]. Consequently, Smarca5-null mice die during early preimplantation due to hypoproliferation of the inner cell mass , whereas Smarca1-null mice develop normally, but show hyperproliferation of progenitors causing an enlarged brain size . Both remodelers have been shown to play a role in the proliferation and differentiation of IPs by controlling FoxG1 expression. FoxG1 is a critical transcription factor for IP proliferation and control of the timing of neurogenesis [269, 270]. On the one hand, conditional knockout of Smarca5 in forebrain progenitors (Emx-cre) resulted in reduced FoxG1 expression, impaired cell cycle kinetics and increased cell death. This resulted in a reduced number of Tbr2+ and FoxG1+ intermediate progenitors and thus a reduced cortical size . Similarly, conditional knockout of Smarca5 (Nestin-cre) caused reduced brain size and cerebral hypoplasia as a result of reduced granular progenitor proliferation . On the other hand, loss of Smarca1 (Ex6DEL) has been shown result in an increased FoxG1 expression, a disruption of progenitor cell cycle kinetics, increased progenitor proliferation and increased neurogenesis . Furthermore, Smarca1 has been shown to directly bind to the promotor of the FoxG1 gene, suggesting that timed chromatin remodelling by SMARCA1 is essential for controlling neuronal development and differentiation . Taken together, these results confirm that timed chromatin remodelling of SWI/SNF-remodelers is essential during neurodevelopment . It is therefore thus not surprising that misexpression of SWI/SNF complex subunits cause NDDs.
SWI/SNF-Related Intellectual Disability Disorders comprise a spectrum of disorders that includes the classic Coffin-Siris syndrome (CSS, OMIM# 135900) and Nicolaides–Baraitser syndrome (NCBRS, OMIM# 601358) . These disorders differ amongst each other in a phenotypic spectrum ranging from syndromic ID over to classic and atypical/severe CSS to NCBRS. The core manifestations of CSS include ID, hypotonia, feeding problems, characteristic facial features, hypertrichosis, sparse scalp hair, visual problems, lax joints, short fifth finger, and one or more underdeveloped nails , and in a minority of the cases, microcephaly . In contrast, NCBRS is defined by ID, short stature, microcephaly, typical face, sparse hair, brachydactyly, prominent interphalangeal joints, behavioural problems, and seizures . Interestingly, structural brain midline defects such as corpus callosum malformations or even absence, are described in the majority of CSS patients [274, 277,278,279,280] and in some NCBRS patients , and animal models for these disorders .
The mild form of CSS is caused by either mutations in the ATPase subunit ARID1B (OMIM# 614556) , or by pathogenic changes in other chromatin remodelling proteins with no direct interaction with SWI/SNF complex, including SOX11 (OMIM# 600898)  and DPF2 (OMIM# 601671) . Additionally, classic and more severe CSS are known to be caused by mutations in SMARCA4 (OMIM# 603254), the common core subunit SMARCB1 (OMIM# 601607), and accessory subunits such as SMARCE1 (OMIM# 603111), ARID1A (OMIM# 603024) and ARID2 (OMIM# 609539) . NCBRS on the contrary has been shown to be caused by mutations in SMARCA2 (OMIM# 600014). Other patients found with mutations in SWI/SNF complex subunits are described for SMARCB1 (OMIM# 601607) have been shown to cause either CSS, but also DOORS syndrome (OMIM# 220500) or Kleefstra syndrome (OMIM# 610253) depending on the site/location of the mutation [286, 287]. Together, these studies on chromatin remodelling complexes support the notion that combinatorial assembly of subunits can instruct cell lineage specification by creating specific patterns of chromatin states at different developmental stages, are essential for normal neurodevelopment [288, 289], and will result in clinical overlapping phenotypes .
Chromatin remodelling complexes: histone variant remodelers
The third class of ATP dependent chromatin remodelers is the INO80 family. The INO80 subfamily is known for its role in histone variant exchange of canonical H2A or H3 histone variants, which is assisted by editing remodelers such as Swr1 complex (SWR1C) , mammalian Snf2-related CBP activator protein (SRCAP)  and p400. Recently, INO80 function in NPCs has been found essential in homologous recombination (HR) DNA repair in a p53-dependent manner . Loss of Ino80 in NPCs (Neurod6Cre/+;Ino80fl/fl) impairs these processes, causing apoptosis and microcephaly in mice . Interestingly, Ino80 deletion in mice leads to unrepaired DNA breaks and apoptosis in symmetric NPC-NPC divisions, but not in asymmetric neurogenic divisions . In correspondence with these findings, INO80 was recently identified as a candidate gene for ID and microcephaly . Among the key histone variants that can be incorporated by the INO80 family is the H2A variant H2A.Z. Specifically, it has been shown that SRCAP removes canonical H2A–H2B dimers and replaces them with H2A.Z–H2B dimers . Little is known yet how mutations in SRCAP affect neurodevelopment. However, mutations in SRCAP have been shown to cause the NDD Floating Harbour Syndrome (OMIM# 136140), which is characterized by intellectual and learning disabilities, a short stature, delayed osseous maturation, language deficits, and distinctive facial features [291, 295,296,297,298].
In addition to the INO80 family, another mechanism for histone exchange is suggested for the α-thalassemia X-linked mental retardation (ATRX) protein. ATRX is an ATP-dependent DNA translocase belonging to the Swi/Snf family of chromatin remodelers . ATRX forms a complex with death domain associated protein (DAXX) [299, 300], and plays a critical role in the replication-independent deposition of the histone variant H3.3, functioning as a histone chaperone at specific genomic regions, including the telomeric domains [301, 302]. Furthermore, ATRX is involved in the suppression of several imprinted genes in the neonatal brain by promoting 3D chromatin structures via CTCF and cohesion . In mice, germline deletion of Atrx has been shown embryonic lethal , whereas conditional deletion of Atrx in NPCs (Foxg1KiCre/+) caused a widespread cellular reduction in both the neocortex and hippocampus resulting in a significant smaller forebrain size . In addition, AtrxFoxg1Cre mice show excessive DNA damage caused by DNA replication stress and subsequent Tp53-dependent apoptosis [306, 307]. On a cellular level, these AtrxFoxg1Cre mice show a reduction in precursor cell number and abnormal migration of progenitors in the hippocampus and the upper layers of the cortex [305, 306]. Furthermore, fewer Neuropeptide Y (NPY), SST and cholecystokinin (CCK) expressing GABAergic neurons were generated in the ventral telencephalon . In humans, mutations in ATRX cause the rare congenital X-linked disorder ATRX syndrome (OMIM# 301040), characterised by moderate to severe ID, DD, microcephaly, hypomyelination, and a mild form of a-thalassemia .
To summarize, chromatin remodelers are highly expressed in neural progenitors, and are essential to dynamically activate, repress, or poise gene expression during the transition from RGCs to glutamatergic neurons or glia (Fig. 2, Table 1). Epigenetic modulation in glutamatergic neuron maturation is reviewed in detail elsewhere , however in the next section we want to highlight the maturation a specialized glutamatergic subpopulation that is frequently impacted in NDDs, called callosal projection neurons.
Epigenetic modulation in callosal projection neuron development
The corpus callosum is a critical link between the two cortical hemispheres. The developmental mechanisms for callosal projections have been well researched in mice (reviewed in [309, 310]), and abnormalities of the corpus callosum often feature in human NDDs, especially ID and epilepsy  but also in Coffin-Siris Syndrome . During mouse brain development, the cortical hemispheres fuse along the midline around E16 , aided by specialized glia populations called midline zipper glia and indusium griseum glia. Those establish the glial wedge to both sides, as well as the bridge-like subcallosal sling. Subsequently, callosal-projecting cortical pyramidal neurons start projecting axons across the midline and connect to their homotopic cortical area. Those projection neurons mostly reside in the upper cortical layers, and their callosal-projecting identity is under direct epigenetic control.
In newly generated pyramidal cell precursors, the transcription factor Ctip2 specifies a subcortical-projecting fate, and is normally expressed in layer 5 neurons. In contrast, in future upper-layer callosal-projecting neurons Ctip2 is repressed by de-acetylation of H4K12 at its promoter region via NuRD complex and HDAC1 recruitment by the DNA-binding protein SATB2 [314,315,316]. After fate specification in the early postnatal period, HDAC1 is gradually removed from the Ctip2 promoter by the transcription factor LMO4, leading to re-establishment of H4K12ac and consequentially re-expression of Ctip2 in a subset of upper-layer neurons [317, 318].
Several other mouse models presented with deficits of callosal projections. For example in Cbp knockout mice (Rubinstein–Taybi Syndrome) a reduced size of the corpus callosum was found , similar to the phenotype of mutants for the chromatin remodelling complex members Prdm8 and HP1γ [320, 321]. Furthermore, knockdown of the chromatin remodeler Chd8 in the neocortex impaired dendrite and axonal growth and branching of upper-layer callosal projection neurons, and resulted in delayed migration of cortical neurons at E18.5, as the majority of labelled cells remained in the VZ/SVZ . Moreover, mutations in the ISWI complex member Smarca5 cause partial agenesis of the corpus callosum, specifically due to reduced generation of viable upper-layer pyramidal neurons during mid-neurogenesis . Lastly, a mouse model for Arid1b and Smarcb1 deficits (Coffin-Siris Syndrome) indeed mirror the human phenotype, as Arid1b+/− and Smarcb1+/inv NesCre+/− mice also have a significantly reduced corpus callosum thickness [277, 282]. Brain-specific Smarcb1+/− mice showed agenesis of the corpus callosum due to midline glia defects, similar to human CSS patients with mutations in SMARCB1, SMARCE1 and ARID1B . In a human patient cohort with non-syndromic callosal abnormalities, mutations in ARID1B were found to be the most common cause, accounting for 10% of all cases .
During the axonal crossover, a multitude of axon guidance factors are required, and defects in the expression of those factors can also cause callosal projection deficits. A recent study described that chromatin remodelling of the axon guidance cue Sema6a caused the callosal defects observed in WAGR Syndrome (OMIM# 194072), a complex disorder including aniridia, kidney tumours, genital abnormalities, and ID . Specifically, this study identified a novel protein, C11orf46/ADP ribosylation factor like GTPase 14 effector protein (ARL14EP), of which mutations were previously associated with ID . C11orf46 is a member of the SETDB1-KRAB associated protein (KAP1)-MCAF1 chromatin repressor complex, and controls H3K9 methylation levels at the Sema6a promoter, cell-autonomously in projection neurons . The callosal projection phenotype could be rescued by targeted H3K9 re-methylation at the Sema6a locus, indicating a direct epigenetic repressive control over axon guidance receptors in callosal-projecting neurons.
Epigenetic modulation in GABAergic neuron development
Besides glutamatergic excitatory neurons, the mammalian neocortex contains between 12 and 20% GABAergic inhibitory neurons [326, 327]. Defects in GABAergic neurons feature prominently in NDDs, in for example epilepsy, Schizophrenia, and ASD (reviewed in ). Broadly, GABAergic neurons can be subdivided according to the expression of marker genes Parvalbumin (PV +), Somatostatin (SST +), and Serotonin Receptor 3a (5-HT3aR +) [329, 330]. Within each of those three large groups, further subdivisions can be made according to gene expression [331, 332], morphology, and electrophysiological firing parameters [333, 334], with current estimates ranging from 20 to 60 subdivisions .
The mechanism of GABAergic neuron generation is comparatively well-conserved between mice and humans [336, 337]. In mice, GABAergic neurons are generated in subdivisions of the Ganglionic Eminences (GEs: Medial GE (MGE), Lateral GE (LGE), Caudal GE (CGE), and Preoptic Area (POA)), which are temporary proliferative zones at the site of the future Striatum. In contrast to the developing cortical plate, the precursors in the GE do form a SVZ, but are not all anchored to the basal membrane, and this population is massively expanded in the human GEs . The precursors divide asymmetrically to produce future GABAergic cells, which gather in the mantle zone and migrate in two morphogen-directed streams towards the cortical plate [338, 339]. The MGE and POA express the transcription factor Nkx2-1 and produce the majority of PV + , and SST + neurons [338, 340] (Fig. 2). In contrast, VIP + neurons (the largest subset of 5-HT3aR + neurons) are produced in the CGE (see for reviews: [338, 341, 342]). Although the networks of transcription factors that define cellular identities during GABAergic neuron development and migration are comparatively well-researched , data on epigenetic regulation and especially chromatin remodelers is scarce and mostly has been inferred from other cell types including cancer biology and other neuron classes .
The first evidence for involvement of chromatin remodelers in GABAergic neuron production was reported in mice with a knockout for the histone acetyltransferase Kat6b/Querkopf, where a reduced density of GAD67 + (GABAergic) neurons in the cortex was found . Years after the initial mouse study, mutations in human KAT6B were found to cause Ohdo/SBBYS syndrome (OMIM# 603736) [345, 346], however the initial findings regarding GABAergic neuron density have not been followed up to date. Mutations in the related KAT6A histone acetyltransferase were also found to cause ID and craniofacial dysmorphism [347, 348], recently described as KAT6A Syndrome . Mouse studies have reproduced a craniofacial phenotype via Hox gene regulation , however neurodevelopmental phenotypes have not yet been studied. We do have a more complete picture for the ATP-dependent chromatin remodeler CHD2, which in humans is associated with epilepsy and broad-spectrum NDDs as described above . Specifically, Chd2 transcription is found to be activated in MGE/POA progenitors by the transcription factor NKX2-1, and by doing so the CHD2 protein in turn colocalizes with NKX2-1 on its downstream targets, illustrating the feedback loops in which chromatin remodelers act . Chd2+/− mice display a marked reduction in MGE-derived GABAergic neuron production, which results in a reduced PV + /SST + GABAergic neuron count in the cortex . The functional consequences (defects in inhibitory synaptic transmission, altered excitatory/inhibitory balance, and behavioural abnormalities) were rescued by an embryonal transplantation of MGE-derived GABAergic neurons, which indicates that already a reduced GABAergic neuron production can produce profound circuit abnormalities .
Haploinsufficiency of the epigenetic regulator ARID1B, which we previously discussed as the causal gene for Coffin-Siris Syndrome, was found to cause premature apoptosis in MGE precursors in mice. As a result, Arid1b+/− mice show a reduced production of MGE-derived (PV + and SST +) GABAergic neurons, and altered laminar arrangement of PV + and SST + neurons in the cortex (Fig. 2) . Mechanistically, the same study found a general reduction of the permissive histone mark H3K9ac3 at the Pvalb promoter in Arid1b+/− mice, resulting in reduced PV transcription throughout development into the adult cortex . Conditional Arid1b knockout in specific GABAergic neuron population showed an interesting bifurcation of effects, as PV-specific Arid1b haploinsufficiency led to reduced mobility and social deficits, while SST-specific Arid1b haploinsufficiency led stereotyped behaviour such as excessive grooming .
Also mutations in the histone acetyltransferase CBP (Rubinstein–Taybi Syndrome), have been implicated in GABAergic precursor generation . Constitutive heterozygous Cbp knockout mice show a transient impairment in GABAergic neuron formation in vivo . Using a more direct approach, region-specific Cbp knockout in the developing MGE reduces the number of PV + and SST + neurons in the cortex and results in a prominent seizure phenotype , indicating that epigenetic regulation by CBP is directly required for proper cell-type specification of inhibitory GABAergic neurons. These results were also confirmed outside of the cortex in non-cortical areas, as conditional knockout of CBP in cerebellar progenitors lead to cerebellar hypoplasia and altered morphology of the cerebellum in both mice (hGFAPCre::CrebbpFl/Fl P25)  and patients . On a cellular level, conditional knockout of CBP in granule cell progenitors altered cerebellar foliation as a result of loss of glial endfeet on the pial surface by Bergmann glia fibers and abnormal Purkinje cells arborisation .
After the formation of GABAergic precursors, the immature GABAergic neurons migrate tangentially following two morphogen-directed streams along the developing cortical plate, where they subsequently invade the cortex in the late stages of corticogenesis between E19 and P4 [341, 360, 361]. The exact place and time for programming the subdivisions within PV + , SST + and 5-HT3aR + is an area of active debate, with different hypotheses highlighting programming at the progenitor stage, during migration to the cortex, or only by local factors in the cortical plate. A recent study indicates that for MGE-derived neurons, the subtype is determined prior to migration, and instructs the migratory route and the place of integration in the cortex . Specifically, the SST + subgroup of Martinotti cells and the PV + subgroup of translaminar PV + neurons preferentially migrate through the Marginal Zone, along the outer side of the developing cortical plate . Migrating GABAergic neurons sense a multitude of environmental cues and integrate them to gene expression patterns. Similarly to GABAergic neuron progenitors, the cascades of transcription factors in migrating GABAergic neurons are comparatively well-characterized, but epigenetic modulations have only recently come into focus (see for review ). A recent series of studies investigated cortical GABAergic neurons derived from the POA, which produces subgroups of SST + , PV + and Reelin + GABAergic neurons . Specifically, POA-derived GABAergic neurons suppress the expression of the transcription factor Pak6 during migration via a non-canonical recruitment of the PRC (specifically EZH2) by DNMT1 to the Pak6 promoter [363, 364]. In POA-specific Dnmt1-knockout mice, the repressive mark H3K27me3 is reduced around the Pak6 transcription start site, leading to precocious expression of Pak6 during migration and consequentially precocious activation of a post-migratory genetic program. As a result, a large portion of POA-derived neurons undergo apoptosis before reaching their cortical destination in POA-specific Dnmt1-knockout mice .
After migrating to the cortex, GABAergic neurons distribute throughout the layers, in a cell-type and area-specific pattern. Broadly speaking, PV + and SST + neurons predominate in the mid- to lower layers, whereas 5-HT3aR + are predominant in layer 1  and (the VIP + subset) in layer 2/3 . Primary sensory areas contain a higher density of PV + neurons, while the areas at the edge of the cortical plate contain a higher density of SST + neurons . Once the GABAergic precursors are located at the appropriate cortical area and laminar location, they integrate into the local circuitry as it develops [367, 368]. The SST + GABAergic neurons mature relatively early, around the same time as the excitatory neurons in the same circuit [362, 369]. However, PV + GABAergic neurons mature much slower, and are dependent on external inputs that activate the local circuit for a successful maturation. The activity levels need to be translated to gene expression patterns, and while no complete mechanism is currently known, the high number of PV + neuron maturation dysfunctions caused by mutations in chromatin remodelers is indicative of a tight epigenetic control over this process . One example is the maturation of PV + neurons in MeCP2+/− mice, which is the primary cause for Rett Syndrome (OMIM# 312750) in humans [371,372,373,374,375]. It is characterized by arrested development between 6 and 18 months of age, regression of acquired skills, loss of speech, stereotypic movements (classically of the hands), microcephaly, seizures, and mental retardation. In MeCP2+/− mice, the lack of MECP2 leads to a premature maturation of PV + cells including marker expression, morphology, and synaptic properties [372, 376]. MECP2 directly binds to the promoter regions of Pvalb and Gad1, the rate-limiting GABA synthesizing enzyme [377, 378]. Also at the adult level, neuronal activity regulates the expression of PV in a dynamic manner , a phenotype which is also impaired in PV + neurons of MeCP2+/− mice , indicating an epigenetic component to the integration of the activity-dependent signal. In contrast, haploinsufficiency of the histone methyltransferase Ehmt1 (Kleefstra Syndrome) causes delayed maturation of PV + neurons in the mouse sensory cortices, consisting of delayed PV expression and PNN generation, as well as reduced GABAergic neurotransmission . Besides neuronal activity levels, PV + neurons also integrate morphogenic signals such as the released transcription factor Otx2. OTX2 is not produced in the cortex, but rather released by thalamic afferents and the Choroid Plexus [382,383,384,385]. OTX2 is taken up by the future PV + neurons, where it upregulates the expression of Gadd45b/g, two DNA demethylases which then mediate the up/downregulation of large sets of genes necessary for the maturation to full PV + cells, including Pvalb itself .
Epigenetic modulation during glia development and function
At the end of the neurogenic period, cortical RGCs cells switch to glial production and generate a vast number of astrocytes and oligodendrocytes . In the mouse cortex, astrocytes are first detected around E16 and oligodendrocytes around birth; however, the vast majority of both cell types are produced during the first month of postnatal development. Cre-loxP lineage tracing showed that oligodendrocytes in the cerebral cortex are produced at different sites outside of the cortex depending on the developmental stage . The first wave of production begins around E12.5 in the MGE and anterior entopeduncular area. The second wave begins around E15 from in the LGE and CGE, and finally, local production begins in the cortical SVZ around birth (Fig. 2) . Similar to oligodendrocytes, astrocytes can be both generated from dividing RGCs , from the postnatal SVZ , or locally by self-amplification in the postnatal cortex  (for detailed information about the origin and specification of glia see [390,391,392,393]). Glia play crucial roles in CNS homeostasis , including synaptic glutamate uptake , synaptogenesis , maintenance of extracellular potassium , nutrient support for neurons , the formation of ECM molecules [399, 400] and many other processes.
Studies investigating the role of chromatin remodelling in mouse models of NDDs have primarily focussed on the alteration of neuronal network function. Recent advances in our understanding of astrocyte function have led to the emerging concept that primary astrocyte dysfunction alone is sufficient to drive the complex behavioural phenotypes observed in some cases of NDDs. As described earlier, RGCs undergo chromatin remodelling in response to various extracellular cues to enable the accessibility of neurogenic or gliogenic genes. Loss of the H3K9 methylase Setdb1 in mice has been shown to reduce H3K9me3 occupancy at the Gfap promotor, resulting in enhanced astrogenesis and accelerated differentiation (Fig. 2) . Furthermore, EHMT1 has been shown to play a role in DNMT1-mediated DNA methylation via UHRF1/LIG1 interaction , which implies that astrocytes might contribute to the neuronal phenotypes in SETDB1-associated disorders or Kleefstra syndrome. Both SETDB1 and EHMT1 have recently been described to coexist in the same complex together with EHMT2 and SUV39H1 , revealing the interesting hypothesis that this complex plays an important role in the neurogenic-to-gliogenic switch, and any dysfunction in any of these genes will lead to a convergent phenotypic outcome.
Epigenetic regulation by the Polycomb Repressor Complex (PRC) has also been described to play a role in the differentiation from NPCs to astrocytes and oligodendrocytes. Acute deletion of the PRC1 component Ring1B or Ezh2 at E12.5 in mice prolonged the neurogenic phase and delayed the astrogenic phase in cultures of neocortical NPCs . In contrast, another report found that cerebral specific loss of Ezh2 in the Emx1-Cre mice accelerated gliogenesis and glial differentiation at P0 . Furthermore, overexpression of Ezh2 in postmitotic astrocytes in turn lead to a downregulation of pro-astrocytic genes S100b and GFAP, whereas an increase in progenitor like genes like SOX2 and CD133 was found . Similarly, a small population of specialized neurogenic astrocytes that resides in the SVZ and survives into adulthood expresses Ezh2, which is required for those astrocytes to keep their neurogenic potential . These results indicate that the PRC associated proteins are essential for promoting the onset of the astrocytic differentiation of NPCs during neocortical development.
Mature glia function has been studied widely in models of NDD (including Noonan syndrome , Neurofibromatosis-1 , Costello syndrome [399, 407], Cardiofaciocutaneous syndrome , Fragile X syndrome , Alexander disease , and Tuberous Sclerosis Complex ) however only in few models of deficient chromatin remodelling. One example is a mouse model for MeCP2 deficiency. Aside from the clear neuronal phenotype found in these mouse models, co-culture studies showed that secreted factors by Mecp2−/− mouse astrocytes significantly affect the development of wild type hippocampal neurons in a non-cell autonomous manner, as was visualised by altered dendrite morphology . Furthermore, neuronal phenotypes found in co-culture with Mecp2−/− astrocytes appear to be dependent upon the expression of astroglial gap-junction protein Connexin-43 (Cx-43), as blocking Cx-43 restored this phenotype . Mecp2−/− mouse astrocytes also showed an increased expression of astroglial marker genes Gfap and S100β and abnormal glutamate clearance . Interestingly, selective restoration of MECP2 in astrocytes in vivo using the Cre-loxP recombination system significantly improves locomotion and anxiety levels, and restores respiratory abnormalities to a normal pattern . At the cellular level, re-expressed MECP2 in astrocytes also restores normal neuronal dendritic morphology . Similar to these findings, an increased expression of GFAP and CX-43 proteins was found in the superior frontal cortex in a cohort of ASD patients . Furthermore, increased levels of H3K9me3 occupancy at the promotor of the gap junction proteins Cx-30 and Cx-43 have been found in cortical and subcortical regions of patients with MDD . This cohort consisted of patients expressing extremely low levels of pro-astrocytic genes GFAP, ALDH1L1, SOX9, GLUL, SCL1A3, GJA1, and GJB6 , proposing a possible role for the H3K9 methylases SETDB1 and SUV39H1 in mature astrocyte function and CX-43 expression .
CHD8 is another example of a chromatin remodeler that plays a role in glia function. Recent studies show cell-type specific Chd8 deletion in OPCs results in myelination defects in mice . In addition to altered myelination, conditional knockout of Chd8 in OPCs (Olig1-Cre;Chd8fl/fl) has been shown to slow down action potential propagation as a result of impaired myelination, leading to deficits including increased social interaction and anxiety-like behaviour as similar to Chd8 heterozygous mutant mice  and behavioural phenotypes found in patients.
Heterozygous loss of Smarca4 (Brg1fl/fl, Nestin-cre) was furthermore shown to cause precocious neuronal differentiation before the onset of gliogenesis . This resulted in a significant reduction of astrocyte and oligodendrocyte differentiation in these animals, suggesting Smarca4 controls the switch from neurogenesis to gliogenesis . Furthermore, SMARCA4 is known as a mediator of long-range interactions of enhancer regions and TTSs , and by doing so is involved in the STAT3 dependent  inter-chromosomal gene clustering of Gfap and Osmr resulting in transcriptional enhancement of these genes . Interestingly, loss of SMARCA2 in SMARCA2K755R/+, and SMARCA2R1159Q/+ NPCs resulted in a reduction of Smarca4 mRNA expression, together with an increased and functionally active binding to chromatin . These results suggested that mutations in SMARCA2 result in global retargeting of SMARCA4, which was shown to drive de novo activation of enhancers and upregulation of astrocyte genes .
To summarize, current evidence shows that chromatin remodelers play a role in the development, migration and circuit integration of each of the major cortical cell classes: Glutamatergic and GABAergic neurons and glia. Consequentially, failures of chromatin remodelling can impact the development of each of those cell types, resulting in a lasting impairment in cellular function.
In this review, we detailed the contribution of chromatin remodelers in different neural cell-classes during the multiple stages of the developmental continuum. Chromatin remodelers are crucial parts of a cell’s information processing machinery, by integrating external and internal signals into gene expression patterns. Developing neurons inhabit an extraordinarily complex epigenetic landscape, and events such as cell-type specification are under tight epigenetic control . Consequentially, defects in chromatin remodelling will lead to a relaxation of that epigenetic control, causing for example premature neural differentiation at the expense of progenitor pool expansion .
As chromatin remodelers have such a variety of functions in different cell types, timepoints and at specific genetic loci, a full picture requires concurrent measurements at several levels simultaneously—a task that current technologies are only starting to address. We see the potential for progress in the following fields:
Understanding chromatin remodeler locus specificity: Chromatin modifications are site-specific on the genome level, such as histone methylation at the activity-dependent Bdnf exon IV . However, until recently, to study this site-specific targeting one had to rely on the cell’s innate targeting abilities. Coupling catalytic subunits to a precise targeting protein allows artificial induction of locus-specific chromatin modifications. One example is the dCas9-SunTAG method [143, 426], where a genetic locus is tagged via dCas9 and gRNAs. Subsequently, local chromatin is modified by a chromatin modifier’s catalytic subunit targeted towards the tag. The ability to induce chromatin modifications at specific genomic sites will improve our understanding of the regulatory networks in gene expression, for example during cell fate specification.
Understanding the role of chromatin remodeler presence in complexes: Chromatin modifiers exist in complexes that dynamically assemble, disassemble, and bind to chromatin at different locations. Complexes are hypothesized to differ between different locations (or time points), however those have proven difficult to investigate with classic immunolabelling techniques. Recent advances in spatial proximity labelling, such as promiscuous biotinylation targeted via dCas9 , allow for a precise snapshot of protein complexes assembled in spatial proximity to a single genomic region. Importantly, this technique can be applied in living cells and in vivo in the developing brain , making it applicable to the neurodevelopmental questions that we have detailed here. This technique allows detailed insights site-specific complex dynamics, a largely unexplored feature of the genetic landscape.
Chromatin remodelling temporal specificity: Neuronal specification is thought to be a series of tightly controlled gene expression (and hence epigenetic regulation) states. For glutamatergic neuron generation in the cortex, recent evidence points to a stochastic generation of different subtypes [121, 429], however it is currently unknown whether GABAergic neuron generation is controlled in a similar way . Classic labelling techniques such as BrdU were only able to identify neurons born within approximately 12 h from each other, which is slower than the hypothesized changes in genetic expression state. Recently developed labelling techniques such as FlashTag selectively label neurons born in a 2-h window in vivo, leading to a more precise identification of the transcriptional program controlling glutamatergic neuron specification [424, 430]. Application of the same technique for GABAergic neurons might deliver interesting insights into subtype specification as well.
Measuring cell-type specificity: The classification of the brain’s cells has been controversial since the start of neuroscience as a field. For example, GABAergic neurons and glia have long resisted simple classification [431, 432]. However, recent large-scale single-cell RNA sequencing studies [331, 332, 433,434,435] attempt to map the cellular diversity of brain from the bottom up. Furthermore, studies measuring multiple modalities on the same neurons promise a unification of classifications from single-cell electrophysiology, morphology and RNA sequencing (Patch-Seq), and have delivered insights in glutamatergic  and GABAergic neuron populations . Especially when coupled with advanced analysis techniques , those large datasets might soon be available as a “reference classifier” that experimental data can be compared with, similarly to reference atlases in neuroanatomy or reference genomes in genomics.
Identification of converging molecular pathways for therapeutic interventions: Functional interactions between several NDD related chromatin remodelers and their regulatory proteins has been shown to converge on a shared transcriptional axis [156, 287, 437,438,439]. One example is the H3K4 demethylase KDM5C, whose expression is controlled by three regulatory proteins: ARX, ZNF711 and PHF8. All four of those genes are located on the X chromosome, and consequentially mutations in any of these four genes are associated with X-linked NDDs [440,441,442]. Interestingly, loss of Arx caused a significant reduction of Kdm5c expression and neuronal maturation in C. Elegans and mice, which could be restored using the HDAC inhibitor SAHA . These findings imply that chromatin remodelers function in closely coupled transcriptional networks, with mutations in genes in the same cluster producing overlapping NDD phenotypes [439, 443]. As demonstrated in the case of KDM5C, overlapping regulatory pathways might be used as drug targets, and mutations in shared pathways could prove to be relatively easily identifiable biomarkers [444, 445]. In a promising first step, several groups have shown that chemical inhibition of HDACs can successfully rescue behavioural phenotypes in mouse models of NDDs [116, 438, 444,445,446,447]. Furthermore, the fact that those pathways tend to be well-conserved might prove valuable in translation to clinical practice.
The studies summarized in this review were almost exclusively performed in animal models of NDDs. While mice have many advantages as model organisms and many features are conserved down to the cellular level , some features appear to be unique to the human lineage, for example specialized cell-types such as outer radial glia cells , subpial interlaminar astrocytes , or the recently described rosehip neurons . Furthermore, some time periods in neurodevelopment are much longer in comparison, elongating the vulnerable periods for many regulatory processes in humans. Therefore, using mouse brains as the sole model we might overlook important human-specific aspects of brain development and NDD pathogenesis. Although the use of animal models will remain essential to study complex developmental processes like cortical layering or migration, human induced pluripotent stem cells (hiPSCs) offer a higher-throughput model to investigate the developmental continuum from the earliest point of progenitor specification until the formation of neuronal circuits in vitro. For this reason, the use hiPSCs has gained a lot of attention recent years in the field of NDD research. HiPSCs provide an unlimited pool of (patient) material, which can be differentiated into neuronal networks, and can be monitored over development in vitro. In addition, these cells are comparatively easy to manipulate using for example CRISPR-Cas9 genome editing, and therefore can be used as a high throughput tool to study genotype–phenotype correlations in a controlled environment [451, 452]. Moreover, patient specific hiPSCs carry the same genetic background as the patient, which allows the study of polygenic disorders like ASD or Schizophrenia that cannot be modelled using animal models.
Protocols for the differentiation of hiPSCs into 3D cerebral organoids are becoming increasingly popular as these models have been shown to resemble the complex developmental programs of early corticogenesis during the first and second trimester of human foetal development [453, 454]. Indeed, 3D cerebral organoids derived from patients with severe microcephaly as a result of CDK5RAP2 mutations showed reduced neuroepithelial differentiation, fewer and smaller progenitor regions, and premature neuronal differentiation . Furthermore, 3D human organoids from idiopathic ASD patients showed reduced proliferation of progenitors, increased neurogenesis, and an imbalance between the production of glutamatergic and GABAergic neurons . Moreover, organoids derived from CNTNAP2+/− hiPSCs showed increased organoid volumes as a result of increased proliferation of progenitors, which in turn expanded the neuronal population . Recent work has shown that patient-derived iPSC organoids with copy number variants in the ASD risk locus 16p11.2 mirror the patient’s micro/macrocephaly phenotype . Similarly, RAB39b loss in 3D organoids has recently been shown to cause hyperproliferation and enlarged organoid size . Studies are currently exploring organoid vascularization to further extend the development and complexity of these organoids [460,461,462], which will allow in the future to study more complex brain phenotypes using these in vitro approaches.
In summary, despite lots of progress in the field, the full influence of chromatin remodelling on neurodevelopment is currently unknown. To fully understand chromatin remodelers’ influence throughout the developmental continuum and identify possible human-specific pathways, future studies should combine human-specific in vitro models such as 3D cerebral organoids and well-characterized developmental models such as mice.
Availability of data and material
Correspondence and should be addressed to firstname.lastname@example.org.
Ernst C (2016) Proliferation and differentiation deficits are a major convergence point for neurodevelopmental disorders. Trends Neurosci 39(5):290–299
Geschwind DH, Flint J (2015) Genetics and genomics of psychiatric disease. Science 349(6255):1489–1494
May PA et al (2018) Prevalence of Fetal Alcohol Spectrum Disorders in 4 US Communities. JAMA 319(5):474–482
Lange S et al (2017) Global Prevalence of Fetal Alcohol Spectrum Disorder Among Children and Youth: A Systematic Review and Meta-analysis. JAMA Pediatrics 171(10):948–956
Kaminen-Ahola N (2020) Fetal alcohol spectrum disorders: Genetic and epigenetic mechanisms. Prenat Diagn 40(9):1185–1192
An JY, Claudianos C (2016) Genetic heterogeneity in autism: From single gene to a pathway perspective. Neurosci Biobehav Rev 68:442–453
Wright CF, FitzPatrick DR, Firth HV (2018) Paediatric genomics: diagnosing rare disease in children. Nat Rev Genet 19(5):253–268
Gilissen C et al (2014) Genome sequencing identifies major causes of severe intellectual disability. Nature 511(7509):344–347
De Rubeis S et al (2014) Synaptic, transcriptional and chromatin genes disrupted in autism. Nature 515(7526):209–215
Iossifov I et al (2014) The contribution of de novo coding mutations to autism spectrum disorder. Nature 515(7526):216–221
Pinto D et al (2014) Convergence of genes and cellular pathways dysregulated in autism spectrum disorders. Am J Hum Genet 94(5):677–694
Satterstrom FK et al (2020) Large-scale exome sequencing study implicates both developmental and functional changes in the neurobiology of autism. Cell 180(3):568–584 (e23)
Ciptasari U, van Bokhoven H (2020) The phenomenal epigenome in neurodevelopmental disorders. Hum Mol Genet 29(R1):R42–R50
Tyagi M et al (2016) Chromatin remodelers: we are the drivers!! Nucleus 7(4):388–404
Hsieh J, Gage FH (2005) Chromatin remodeling in neural development and plasticity. Curr Opin Cell Biol 17(6):664–671
Lomvardas S, Maniatis T (2016) Histone and DNA modifications as regulators of neuronal development and function. Cold Spring Harb Perspect Biol 8(7):a024208
Gabriele M et al (2018) The chromatin basis of neurodevelopmental disorders: rethinking dysfunction along the molecular and temporal axes. Prog Neuropsychopharmacol Biol Psychiatry 84:306–327
Parenti I et al (2020) Neurodevelopmental disorders: from genetics to functional pathways. Trends Neurosci 43(8):608–621
Iwase S et al (2017) Epigenetic etiology of intellectual disability. J Neurosci 37(45):10773–10782
De Majo F, Calore M (2018) Chromatin remodelling and epigenetic state regulation by non-coding RNAs in the diseased heart. Non-coding RNA Res 3(1):20–28
Böhmdorfer G, Wierzbicki AT (2015) Control of chromatin structure by long noncoding RNA. Trends Cell Biol 25(10):623–632
Wei JW et al (2017) Non-coding RNAs as regulators in epigenetics (Review). Oncol Rep 37(1):3–9
Flemming W (1882) Zellsubstanz, Kern und Zelltheilung, ed. F.C.W. Vogel. Leipzig.
Luger K et al (1997) Characterization of nucleosome core particles containing histone proteins made in bacteria. J Mol Biol 272(3):301–311
Heitz E (1928) Das Heterochromatin der Moose. Bornträger.
Sadakierska-Chudy A, Filip M (2015) A comprehensive view of the epigenetic landscape Part II: Histone post-translational modification, nucleosome level, and chromatin regulation by ncRNAs. Neurotox Res 27(2):172–197
Clapier CR, Cairns BR (2009) The biology of chromatin remodeling complexes. Annu Rev Biochem 78:273–304
Davis L, Onn I, Elliott E (2018) The emerging roles for the chromatin structure regulators CTCF and cohesin in neurodevelopment and behavior. Cell Mol Life Sci 75(7):1205–1214
Zhang Y, Reinberg D (2001) Transcription regulation by histone methylation: interplay between different covalent modifications of the core histone tails. Genes Dev 15(18):2343–2360
Sterner DE, Berger SL (2000) Acetylation of histones and transcription-related factors. Microbiol Mol Biol Rev 64(2):435–459
Nowak SJ, Corces VG (2004) Phosphorylation of histone H3: a balancing act between chromosome condensation and transcriptional activation. Trends Genet 20(4):214–220
Shilatifard A (2006) Chromatin modifications by methylation and ubiquitination: implications in the regulation of gene expression. Annu Rev Biochem 75:243–269
Nathan D et al (2006) Histone sumoylation is a negative regulator in Saccharomyces cerevisiae and shows dynamic interplay with positive-acting histone modifications. Genes Dev 20(8):966–976
Hymes J, Fleischhauer K, Wolf B (1995) Biotinylation of biotinidase following incubation with biocytin. Clin Chim Acta 233(1–2):39–45
Hassa PO et al (2006) Nuclear ADP-ribosylation reactions in mammalian cells: where are we today and where are we going? Microbiol Mol Biol Rev 70(3):789–829
Cuthbert GL et al (2004) Histone deimination antagonizes arginine methylation. Cell 118(5):545–553
Wang Y et al (2004) Human PAD4 regulates histone arginine methylation levels via demethylimination. Science 306(5694):279–283
Nelson CJ, Santos-Rosa H, Kouzarides T (2006) Proline isomerization of histone H3 regulates lysine methylation and gene expression. Cell 126(5):905–916
Kim E et al (1997) Deficiency of a protein-repair enzyme results in the accumulation of altered proteins, retardation of growth, and fatal seizures in mice. Proc Natl Acad Sci USA 94(12):6132–6137
Tan M et al (2011) Identification of 67 histone marks and histone lysine crotonylation as a new type of histone modification. Cell 146(6):1016–1028
Chen Y et al (2007) Lysine propionylation and butyrylation are novel post-translational modifications in histones. Mol Cell Proteomics 6(5):812–819
Farrelly LA et al (2019) Histone serotonylation is a permissive modification that enhances TFIID binding to H3K4me3. Nature 567(7749):535–539
Lepack AE et al (2020) Dopaminylation of histone H3 in ventral tegmental area regulates cocaine seeking. Science 368(6487):197–201
Tan M et al (2014) Lysine glutarylation is a protein posttranslational modification regulated by SIRT5. Cell Metab 19(4):605–617
Wagner GR et al (2017) A class of reactive Acyl-CoA species reveals the non-enzymatic origins of protein acylation. Cell Metab 25(4):823-837.e8
Bao X et al (2019) Glutarylation of histone H4 Lysine 91 regulates chromatin dynamics. Mol Cell 76(4):660-675.e9
Zhang D et al (2019) Metabolic regulation of gene expression by histone lactylation. Nature 574(7779):575–580
Huang H et al (2018) Lysine benzoylation is a histone mark regulated by SIRT2. Nat Commun 9(1):3374
Wilson JP et al (2011) Proteomic analysis of fatty-acylated proteins in mammalian cells with chemical reporters reveals S-acylation of histone H3 variants. Mol Cell Proteom 10(3):110001198
Zou C et al (2011) Acyl-CoA:lysophosphatidylcholine acyltransferase I (Lpcat1) catalyzes histone protein O-palmitoylation to regulate mRNA synthesis. J Biol Chem 286(32):28019–28025
Webby CJ et al (2009) Jmjd6 catalyses lysyl-hydroxylation of U2AF65, a protein associated with RNA splicing. Science 325(5936):90–93
Zheng Q et al (2019) Reversible histone glycation is associated with disease-related changes in chromatin architecture. Nat Commun 10(1):1289
Galligan JJ et al (2018) Methylglyoxal-derived posttranslational arginine modifications are abundant histone marks. Proc Natl Acad Sci U S A 115(37):9228–9233
Galligan JJ et al (2014) Stable histone adduction by 4-oxo-2-nonenal: a potential link between oxidative stress and epigenetics. J Am Chem Soc 136(34):11864–11866
Jin J et al (2016) SIRT2 Reverses 4-Oxononanoyl Lysine Modification on Histones. J Am Chem Soc 138(38):12304–12307
Chen D et al (2013) Cigarette smoke component acrolein modulates chromatin assembly by inhibiting histone acetylation. J Biol Chem 288(30):21678–21687
Fang L et al (2016) Mechanisms underlying acrolein-mediated inhibition of chromatin assembly. Mol Cell Biol 36(23):2995–3008
Xu L et al (2015) Crosstalk of homocysteinylation, methylation and acetylation on histone H3. Analyst 140(9):3057–3063
Zhang Q et al (2018) Elevated H3K79 homocysteinylation causes abnormal gene expression during neural development and subsequent neural tube defects. Nat Commun 9(1):3436
Bustin M (1971) Nitration of the tyrosine in histone F1 in salt solutions and in F1-polyanion complexes. Biochim Biophys Acta 251(2):172–180
Prütz WA et al (1985) Reactions of nitrogen dioxide in aqueous model systems: oxidation of tyrosine units in peptides and proteins. Arch Biochem Biophys 243(1):125–134
Haqqani AS, Kelly JF, Birnboim HC (2002) Selective nitration of histone tyrosine residues in vivo in mutatect tumors. J Biol Chem 277(5):3614–3621
Gould N et al (2013) Regulation of protein function and signaling by reversible cysteine S-nitrosylation. J Biol Chem 288(37):26473–26479
Lee CF, Paull TT, Person MD (2013) Proteome-wide detection and quantitative analysis of irreversible cysteine oxidation using long column UPLC-pSRM. J Proteome Res 12(10):4302–4315
Akter S et al (2018) Chemical proteomics reveals new targets of cysteine sulfinic acid reductase. Nat Chem Biol 14(11):995–1004
García-Giménez JL et al (2013) Histone h3 glutathionylation in proliferating mammalian cells destabilizes nucleosomal structure. Antioxid Redox Signal 19(12):1305–1320
Zhou M, Paša-Tolić L, Stenoien DL (2017) Profiling of histone post-translational modifications in mouse brain with high-resolution top-down mass spectrometry. J Proteome Res 16(2):599–608
Bjornsson HT (2015a) The Mendelian disorders of the epigenetic machinery. Genome Res 25(10):1473–1481
Chandy M et al (2006) SWI/SNF displaces SAGA-acetylated nucleosomes. Eukaryot Cell 5(10):1738–1747
Hassan YI, Zempleni J (2006) Epigenetic regulation of chromatin structure and gene function by biotin. J Nutr 136(7):1763–1765
Ito T et al (2000) p300-mediated acetylation facilitates the transfer of histone H2A–H2B dimers from nucleosomes to a histone chaperone. Genes Dev 14(15):1899–1907
Reinke H, Horz W (2003) Histones are first hyperacetylated and then lose contact with the activated PHO5 promoter. Mol Cell 11(6):1599–1607
Bowman GD, Poirier MG (2015) Post-translational modifications of histones that influence nucleosome dynamics. Chem Rev 115(6):2274–2295
Ruthenburg AJ et al (2007) Multivalent engagement of chromatin modifications by linked binding modules. Nat Rev Mol Cell Biol 8(12):983–994
Wang Z, Patel DJ (2011) Combinatorial readout of dual histone modifications by paired chromatin-associated modules. J Biol Chem 286(21):18363–18368
Vallianatos CN et al (2019) Amelioration of brain histone methylopathies by balancing a Writer-Eraser Duo KMT2A-KDM5C. bioRxiv 2019:567917
Yu N-K, Baek SH, Kaang B-K (2011) DNA methylation-mediated control of learning and memory. Mol Brain 4(1):5
Okano M et al (1999) DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 99(3):247–257
Bestor TH (2000) The DNA methyltransferases of mammals. Hum Mol Genet 9(16):2395–2402
Jones PL et al (1998) Methylated DNA and MeCP2 recruit histone deacetylase to repress transcription. Nat Genet 19(2):187–191
Nan X et al (1998) Transcriptional repression by the methyl-CpG-binding protein MeCP2 involves a histone deacetylase complex. Nature 393(6683):386–389
Williams SR et al (2010) Haploinsufficiency of MBD5 associated with a syndrome involving microcephaly, intellectual disabilities, severe speech impairment, and seizures. Eur J Hum Genet 18(4):436–441
Talkowski ME et al (2011) Assessment of 2q23.1 microdeletion syndrome implicates MBD5 as a single causal locus of intellectual disability, epilepsy, and autism spectrum disorder. Am J Human Genet 89(4):551–563
Bird A (2008) The methyl-CpG-binding protein MeCP2 and neurological disease. Biochem Soc Trans 36(Pt 4):575–583
Zeng Y, Chen T (2019) DNA methylation reprogramming during mammalian development. Genes 10(4):257
Lee TW, Katz DJ (2020) Hansel, gretel, and the consequences of failing to remove histone methylation breadcrumbs. Trends Genet 36(3):160–176
Greenberg MVC, Bourc’his D (2019) The diverse roles of DNA methylation in mammalian development and disease. Nat Rev Mol Cell Biol 20(10):590–607
Cerrato F et al (2020) DNA methylation in the diagnosis of monogenic diseases. Genes (Basel) 11(4):355
Sadikovic B et al (2019) DNA methylation signatures in mendelian developmental disorders as a diagnostic bridge between genotype and phenotype. Epigenomics 11(5):563–575
Schenkel LC et al (2016) DNA methylation analysis in constitutional disorders: clinical implications of the epigenome. Crit Rev Clin Lab Sci 53(3):147–165
Sadikovic B, Levy MA, Aref-Eshghi E (2020) Functional annotation of genomic variation: DNA methylation episignatures in neurodevelopmental Mendelian disorders. Hum Mol Genet 29(R1):R27–R32
Reilly J, Kerkhof J, Sadikovic B (2020) EpiSigns: DNA methylation signatures in mendelian neurodevelopmental disorders as a diagnostic link between a genotype and phenotype. Adv Mol Pathol 3:29–39
Aref-Eshghi E et al (2020) Evaluation of DNA methylation episignatures for diagnosis and phenotype correlations in 42 Mendelian neurodevelopmental disorders. Am J Human Genet 106(3):356–370
Aref-Eshghi E et al (2019) Diagnostic utility of genome-wide dna methylation testing in genetically unsolved individuals with suspected hereditary conditions. Am J Human Genet 104(4):685–700
Aref-Eshghi E et al (2018) Genomic DNA methylation signatures enable concurrent diagnosis and clinical genetic variant classification in neurodevelopmental syndromes. Am J Hum Genet 102(1):156–174
Krogan NJ et al (2003) A Snf2 family ATPase complex required for recruitment of the histone H2A variant Htz1. Mol Cell 12(6):1565–1576
Murawska M, Brehm A (2011) CHD chromatin remodelers and the transcription cycle. Transcription 2(6):244–253
Cremer T, Cremer C (2001) Chromosome territories, nuclear architecture and gene regulation in mammalian cells. Nat Rev Genet 2(4):292–301
Lieberman-Aiden E et al (2009) Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326(5950):289–293
Dixon JR et al (2012) Topological domains in mammalian genomes identified by analysis of chromatin interactions. Nature 485(7398):376–380
de Laat W, Duboule D (2013) Topology of mammalian developmental enhancers and their regulatory landscapes. Nature 502(7472):499–506
Phillips-Cremins JE et al (2013) Architectural protein subclasses shape 3D organization of genomes during lineage commitment. Cell 153(6):1281–1295
Suhas Rao SP et al (2014) A 3D map of the human genome at kilobase resolution reveals principles of chromatin looping. Cell 159(7):1665–1680
Xiao T, Wallace J, Felsenfeld G (2011) Specific sites in the C terminus of CTCF interact with the SA2 subunit of the cohesin complex and are required for cohesin-dependent insulation activity. Mol Cell Biol 31(11):2174–2183
Vietri Rudan M et al (2015) Comparative Hi-C reveals that CTCF underlies evolution of chromosomal domain architecture. Cell Reports 10(8):1297–1309
Schwalie PC et al (2013) Co-binding by YY1 identifies the transcriptionally active, highly conserved set of CTCF-bound regions in primate genomes. Genome Biol 14(12):R148
Mourad R, Cuvier O (2016) Computational identification of genomic features that influence 3D chromatin domain formation. PLoS Comput Biol 12(5):e1004908
Cournac A, Koszul R, Mozziconacci J (2015) The 3D folding of metazoan genomes correlates with the association of similar repetitive elements. Nucleic Acids Res 44(1):245–255
Melé M, John L (2016) Rinn, “Cat’s cradling” the 3D genome by the Act of LncRNA transcription. Mol Cell 62(5):657–664
Gorkin DU, Leung D, Ren B (2014) The 3D genome in transcriptional regulation and pluripotency. Cell Stem Cell 14(6):762–775
Bastle RM, Maze I (2019) Chromatin regulation in complex brain disorders. Curr Opin Behav Sci 25:57–65
Boukas L et al (2019) Coexpression patterns define epigenetic regulators associated with neurological dysfunction. Genome Res 29(4):532–542
Fahrner JA, Bjornsson HT (2019) Mendelian disorders of the epigenetic machinery: postnatal malleability and therapeutic prospects. Hum Mol Genet 28:R254–R264
Iwase S, Martin DM (2018) Chromatin in nervous system development and disease. Mol Cell Neurosci 87:1–3
Woodworth MB et al (2012) SnapShot: cortical development. Cell 151(4):918-918.e1
Greig LC et al (2013) Molecular logic of neocortical projection neuron specification, development and diversity. Nat Rev Neurosci 14(11):755–769
Nowakowski TJ et al (2016) Transformation of the radial glia scaffold demarcates two stages of human cerebral cortex development. Neuron 91(6):1219–1227
Ge W-P et al (2012) Local generation of glia is a major astrocyte source in postnatal cortex. Nature 484(7394):376–380
deAzevedo LC et al (2003) Cortical radial glial cells in human fetuses: Depth-correlated transformation into astrocytes. J Neurobiol 55(3):288–298
Kadhim HJ, Gadisseux JF, Evrard P (1988) Topographical and cytological evolution of the glial phase during prenatal development of the human brain: histochemical and electron microscopic study. J Neuropathol Exp Neurol 47(2):166–188
Llorca A et al (2019) A stochastic framework of neurogenesis underlies the assembly of neocortical cytoarchitecture. eLife 8:e51381
Awad S et al (2013) Mutation in PHC1 implicates chromatin remodeling in primary microcephaly pathogenesis. Hum Mol Genet 22(11):2200–2213
Beunders G et al (2013) Exonic deletions in AUTS2 cause a syndromic form of intellectual disability and suggest a critical role for the C terminus. Am J Hum Genet 92(2):210–220
Nagamani SCS et al (2013) Detection of copy-number variation in AUTS2 gene by targeted exonic array CGH in patients with developmental delay and autistic spectrum disorders. Eur J Hum Genet 21(3):343–346
Margueron R et al (2008) Ezh1 and Ezh2 maintain repressive chromatin through different mechanisms. Mol Cell 32(4):503–518
Weaver DD et al (1974) A new overgrowth syndrome with accelerated skeletal maturation, unusual facies, and camptodactyly. J Pediatr 84(4):547–552
Gibson WT et al (2012) Mutations in EZH2 cause weaver syndrome. Am J Hum Genet 90(1):110–118
Choufani S et al (2020) DNA methylation signature for EZH2 functionally classifies sequence variants in three PRC2 complex genes. Am J Hum Genet 106(5):596–610
Pereira JD et al (2010) Ezh2, the histone methyltransferase of PRC2, regulates the balance between self-renewal and differentiation in the cerebral cortex. Proc Natl Acad Sci 107(36):15957–15962
Telley L et al (2019) Temporal patterning of apical progenitors and their daughter neurons in the developing neocortex. Science 364(6440):eaav522
Oberst P et al (2019) Temporal plasticity of apical progenitors in the developing mouse neocortex. Nature 573(7774):370–374
Molyneaux BJ et al (2007) Neuronal subtype specification in the cerebral cortex. Nat Rev Neurosci 8(6):427–437
Gao P et al (2014) Deterministic progenitor behavior and unitary production of neurons in the neocortex. Cell 159(4):775–788
Donega V et al (2018) Transcriptional dysregulation in postnatal glutamatergic progenitors contributes to closure of the cortical neurogenic period. Cell Rep 22(10):2567–2574
Amberg N, Laukoter S, Hippenmeyer S (2019) Epigenetic cues modulating the generation of cell-type diversity in the cerebral cortex. J Neurochem 149(1):12–26
Hirabayashi Y et al (2009) Polycomb limits the neurogenic competence of neural precursor cells to promote astrogenic fate transition. Neuron 63(5):600–613
Sparmann A et al (2013) The chromodomain helicase Chd4 is required for Polycomb-mediated inhibition of astroglial differentiation. Embo j 32(11):1598–1612
Zhao L et al (2015) Ezh2 is involved in radial neuronal migration through regulating Reelin expression in cerebral cortex. Sci Rep 5(1):15484
Morimoto-Suzki N et al (2014) The polycomb component Ring1B regulates the timed termination of subcerebral projection neuron production during mouse neocortical development. Development 141(22):4343–4353
Cukier HN et al (2012) The expanding role of MBD genes in autism: identification of a MECP2 duplication and novel alterations in MBD5, MBD6, and SETDB1. Autism Res 5(6):385–397
Xu Q et al (2016) Chromosomal microarray analysis in clinical evaluation of neurodevelopmental disorders-reporting a novel deletion of SETDB1 and illustration of counseling challenge. Pediatr Res 80(3):371–381
Jiang Y et al (2010) Setdb1 histone methyltransferase regulates mood-related behaviors and expression of the NMDA receptor subunit NR2B. J Neurosci 30(21):7152–7167
Jiang Y et al (2017) The methyltransferase SETDB1 regulates a large neuron-specific topological chromatin domain. Nat Genet 49(8):1239–1250
Ripke S et al (2014) Biological insights from 108 schizophrenia-associated genetic loci. Nature 511(7510):421–427
Cruvinel E et al (2014) Reactivation of maternal SNORD116 cluster via SETDB1 knockdown in Prader-Willi syndrome iPSCs. Hum Mol Genet 23(17):4674–4685
Zhu Y et al (2020) Epigenetic mechanism of SETDB1 in brain: implications for neuropsychiatric disorders. Transl Psychiatry 10(1):115
Ling BM et al (2012) Lysine methyltransferase G9a methylates the transcription factor MyoD and regulates skeletal muscle differentiation. Proc Natl Acad Sci USA 109(3):841–846
Sripathy SP, Stevens J, Schultz DC (2006) The KAP1 corepressor functions to coordinate the assembly of De Novo HP1-demarcated microenvironments of heterochromatin required for KRAB zinc finger protein-mediated transcriptional repression. Mol Cell Biol 26(22):8623–8638
Schultz DC, Friedman JR, Rauscher FJ 3rd (2001) Targeting histone deacetylase complexes via KRAB-zinc finger proteins: the PHD and bromodomains of KAP-1 form a cooperative unit that recruits a novel isoform of the Mi-2alpha subunit of NuRD. Genes Dev 15(4):428–443
Timms RT et al (2016) ATF7IP-mediated stabilization of the histone methyltransferase SETDB1 is essential for heterochromatin formation by the HUSH complex. Cell Rep 17(3):653–659
Minkovsky A et al (2014) The Mbd1-Atf7ip-Setdb1 pathway contributes to the maintenance of X chromosome inactivation. Epigenet Chromatin 7(1):12
Tian Z et al (2006) Expression of DNA methyltransferases in salivary adenoid cystic carcinoma and its association with the CpG islands methylation of tumor suppressor genes. Zhonghua Kou Qiang Yi Xue Za Zhi 41(7):411–415
Tachibana M et al (2002) G9a histone methyltransferase plays a dominant role in euchromatic histone H3 lysine 9 methylation and is essential for early embryogenesis. Genes Dev 16(14):1779–1791
Chen X et al (2012) G9a/GLP-dependent histone H3K9me2 patterning during human hematopoietic stem cell lineage commitment. Genes Dev 26(22):2499–2511
Wen B et al (2009) Large histone H3 lysine 9 dimethylated chromatin blocks distinguish differentiated from embryonic stem cells. Nat Genet 41(2):246–250
Chen ES et al (2014) Molecular convergence of neurodevelopmental disorders. Am J Hum Genet 95(5):490–508
Olsen JB et al (2016) G9a and ZNF644 physically associate to suppress progenitor gene expression during neurogenesis. Stem Cell Rep 7(3):454–470
Schaefer A et al (2009) Control of cognition and adaptive behavior by the GLP/G9a epigenetic suppressor complex. Neuron 64(5):678–691
Balemans MCM et al (2014) Reduced euchromatin histone methyltransferase 1 causes developmental delay, hypotonia, and cranial abnormalities associated with increased bone gene expression in Kleefstra syndrome mice. Dev Biol 386(2):395–407
Willemsen MH et al (2012) Update on kleefstra syndrome. Mol Syndromol 2(3–5):202–212
Vermeulen K et al (2017) Adaptive and maladaptive functioning in Kleefstra syndrome compared to other rare genetic disorders with intellectual disabilities. Am J Med Genet A 173(7):1821–1830
Wang J et al (2010) CBP histone acetyltransferase activity regulates embryonic neural differentiation in the normal and Rubinstein-Taybi syndrome brain. Dev Cell 18(1):114–125
Yao T-P et al (1998) Gene dosage-dependent embryonic development and proliferation defects in mice lacking the transcriptional integrator p300. Cell 93(3):361–372
Lipinski M, Del Blanco B, Barco A (2019) CBP/p300 in brain development and plasticity: disentangling the KAT’s cradle. Curr Opin Neurobiol 59:1–8
Ateca-Cabarga JC et al (2015) Brain size regulations by cbp haploinsufficiency evaluated by in-vivo MRI based volumetry. Sci Rep 5:16256
Li L et al (2020) Lysine acetyltransferase 8 is involved in cerebral development and syndromic intellectual disability. J Clin Invest 130(3):1431–1445
Koolen DA et al (2012) Mutations in the chromatin modifier gene KANSL1 cause the 17q21.31 microdeletion syndrome. Nat Genet 44(6):639–641
Arbogast T et al (2017) Mouse models of 17q21.31 microdeletion and microduplication syndromes highlight the importance of Kansl1 for cognition. PLoS Genet 13(7):1006886
D’Mello SR (2019) Regulation of central nervous system development by class I histone deacetylases. Dev Neurosci 41(3):149–165
Hsieh J, Gage FH (2004) Epigenetic control of neural stem cell fate. Curr Opin Genet Dev 14(5):461–469
de Ruijter AJ et al (2003) Histone deacetylases (HDACs): characterization of the classical HDAC family. Biochem J 370(Pt 3):737–749
MacDonald JL, Roskams AJ (2008) Histone deacetylases 1 and 2 are expressed at distinct stages of neuro-glial development. Dev Dyn 237(8):2256–2267
Montgomery RL et al (2009) Histone deacetylases 1 and 2 control the progression of neural precursors to neurons during brain development. Proc Natl Acad Sci USA 106(19):7876–7881
Hagelkruys A et al (2014) A single allele of Hdac2 but not Hdac1 is sufficient for normal mouse brain development in the absence of its paralog. Development 141(3):604–616
Yuniarti N et al (2013) Prenatal exposure to suberoylanilide hydroxamic acid perturbs corticogenesis. Neurosci Res 77(1–2):42–49
Ye F et al (2009) HDAC1 and HDAC2 regulate oligodendrocyte differentiation by disrupting the beta-catenin-TCF interaction. Nat Neurosci 12(7):829–838
Wagner VF et al (2019) A De novo HDAC2 variant in a patient with features consistent with Cornelia de Lange syndrome phenotype. Am J Med Genet A 179(5):852–856
Gao X et al (2018) A functional mutation in HDAC8 gene as novel diagnostic marker for cornelia de lange syndrome. Cell Physiol Biochem 47(6):2388–2395
Helgeson M et al (2018) Molecular characterization of HDAC8 deletions in individuals with atypical Cornelia de Lange syndrome. J Hum Genet 63(3):349–356
Deardorff MA, Porter NJ, Christianson DW (2016) Structural aspects of HDAC8 mechanism and dysfunction in Cornelia de Lange syndrome spectrum disorders. Protein Sci 25(11):1965–1976
Deardorff MA et al (2012a) HDAC8 mutations in cornelia de lange syndrome affect the cohesin acetylation cycle. Nature 489(7415):313–317
Bottai D et al (2018) Modeling Cornelia de Lange syndrome in vitro and in vivo reveals a role for cohesin complex in neuronal survival and differentiation. Hum Mol Genet 28(1):64–73
Katayama S et al (2018) HDAC8 regulates neural differentiation through embryoid body formation in P19cells. Biochem Biophys Res Commun 498(1):45–51
Harakalova M et al (2012) X-exome sequencing identifies a HDAC8 variant in a large pedigree with X-linked intellectual disability, truncal obesity, gynaecomastia, hypogonadism and unusual face. J Med Genet 49(8):539–543
Aoto T et al (2006) Nuclear and chromatin reorganization in the MHC-Oct3/4 locus at developmental phases of embryonic stem cell differentiation. Dev Biol 298(2):354–367
Gregor A et al (2013) De novo mutations in the genome organizer CTCF cause intellectual disability. Am J Hum Genet 93(1):124–131
Bastaki F et al (2017) Identification of a novel CTCF mutation responsible for syndromic intellectual disability—a case report. BMC Med Genet 18(1):68–68
Konrad EDH et al (2019) CTCF variants in 39individuals with a variable neurodevelopmental disorder broaden the mutational andclinical spectrum. Genet Med 21(12):2723–2733
Watson LA et al (2014) Dual effect of CTCF loss on neuroprogenitor differentiation and survival. J Neurosci 34(8):2860–2870
Hirayama T et al (2012) CTCF is required for neural development and stochastic expression of clustered Pcdh genes in neurons. Cell Rep 2(2):345–357
Deardorff MA et al (2007) Mutations in cohesin complex members SMC3 and SMC1A cause a mild variant of cornelia de Lange syndrome with predominant mental retardation. Am J Hum Genet 80(3):485–494
Deardorff MA et al (2012b) RAD21 mutations cause a human cohesinopathy. Am J Hum Genet 90(6):1014–1027
Krantz ID et al (2004) Cornelia de Lange syndrome is caused by mutations in NIPBL, the human homolog of Drosophila melanogaster Nipped-B. Nat Genet 36(6):631–635
Vega H et al (2005) Roberts syndrome is caused by mutations in ESCO2, a human homolog of yeast ECO1 that is essential for the establishment of sister chromatid cohesion. Nat Genet 37(5):468–470
White JK et al (2013) Genome-wide Generation and Systematic Phenotyping of Knockout Mice Reveals New Roles for Many Genes. Cell 154(2):452–464
Lai AY, Wade PA (2011) Cancer biology and NuRD: a multifaceted chromatin remodelling complex. Nat Rev Cancer 11(8):588–596
Woodage T et al (1997) Characterization of the CHD family of proteins. Proc Natl Acad Sci USA 94(21):11472–11477
Guzman-Ayala M et al (2015) Chd1 is essential for the high transcriptional output and rapid growth of the mouse epiblast. Development 142(1):118–127
Gaspar-Maia A et al (2009) Chd1 regulates open chromatin and pluripotency of embryonic stem cells. Nature 460(7257):863–868
Pilarowski GO et al (2018) Missense variants in the chromatin remodeler CHD1 are associated with neurodevelopmental disability. J Med Genet 55(8):561–566
Shen T et al (2015) CHD2 is required for embryonic neurogenesis in the developing cerebral cortex. Stem Cells 33(6):1794–1806
Kim YJ et al (2018) Chd2 is necessary for neural circuit development and long-term memory. Neuron 0:1–14
Homem CC, Repic M, Knoblich JA (2015) Proliferation control in neural stem and progenitor cells. Nat Rev Neurosci 16(11):647–659
Thomas RH et al (2015) CHD2 myoclonic encephalopathy is frequently associated with self-induced seizures. Neurology 84(9):951–958
Study, D.D.D. (2017) Prevalence and architecture of de novo mutations in developmental disorders. Nature 542(7642):433–438
Chénier S et al (2014) CHD2 haploinsufficiency is associated with developmental delay, intellectual disability, epilepsy and neurobehavioural problems. J Neurodev Disord 6(1):9
Lamar KJ, Carvill GL (2018) Chromatin remodeling proteins in epilepsy: lessons from CHD2-associated epilepsy. Front Mol Neurosci 11:208
Nitarska J et al (2016) A functional switch of NuRD chromatin remodeling complex subunits regulates mouse cortical development. Cell Rep 17(6):1683–1698
Snijders Blok L et al (2018) CHD3 helicase domain mutations cause a neurodevelopmental syndrome with macrocephaly and impaired speech and language. Nat Commun 9(1):4619
Drivas TG et al (2020) A second cohort of CHD3 patients expands the molecular mechanisms known to cause Snijders Blok-Campeau syndrome. Eur J Hum Genet 28(10):1422–1431
Weiss K et al (2016) De novo mutations in CHD4, an ATP-dependent chromatin remodeler gene, cause an intellectual disability syndrome with distinctive dysmorphisms. Am J Hum Genet 99(4):934–941
Weiss K et al (2020) The CHD4-related syndrome: a comprehensive investigation of theclinical spectrum, genotype–phenotype correlations, and molecular basis. Genet Med 22(2):389–397
Kovač K et al (2018) Tumour-associated missense mutations in the dMi-2 ATPase alters nucleosome remodelling properties in a mutation-specific manner. Nat Commun 9(1):2112
Ragheb R et al (2020) Differential regulation of lineage commitment in human and mouse primed pluripotent stem cells by NuRD. bioRxiv 2020:2020.02.05.935544
Egan CM et al (2013) CHD5 is required for neurogenesis and has a dual role in facilitating gene expression and polycomb gene repression. Dev Cell 26(3):223–236
Thompson PM et al (2003) CHD5, a new member of the chromodomain gene family, is preferentially expressed in the nervous system. Oncogene 22(7):1002–1011
Potts RC et al (2011) CHD5, a brain-specific paralog of Mi2 chromatin remodeling enzymes, regulates expression of neuronal genes. PLoS ONE 6(9):e24515
Pisansky MT et al (2017) Mice lacking the chromodomain helicase DNA-binding 5 chromatin remodeler display autism-like characteristics. Transl Psychiatry 7(6):e1152
Mills AA (2017) The chromodomain helicase DNA-binding chromatin remodelers: family traits that protect from and promote cancer. Cold Spring Harb Perspect Med 7(4):a026450
Kalscheuer VM et al (2008) Disruption of the TCF4 gene in a girl with mental retardation but without the classical Pitt-Hopkins syndrome. Am J Med Genet A 146A(16):2053–2059
Yamada K et al (2010) Characterization of a de novo balanced t(4;20)(q33;q12) translocation in a patient with mental retardation. Am J Med Genet A 152A(12):3057–3067
Douet-Guilbert N et al (2015) A novel translocation (6;20)(q13;q12) in acute myeloid leukemia likely results in LMBRD1–CHD6 fusion. Leukemia Lymphoma 56(2):527–528
Kargapolova Y et al (2020) Overarching control of autophagy and DNA damage response by CHD6 revealed by modeling a rare human pathology. bioRxiv 2020:2020.01.27.921171
Hurd EA et al (2007) Loss of Chd7 function in gene-trapped reporter mice is embryonic lethal and associated with severe defects in multiple developing tissues. Mamm Genome 18(2):94–104
Yao H et al (2020) CHD7 promotes neural progenitor differentiation in embryonic stem cells via altered chromatin accessibility and nascent gene expression. Sci Rep 10(1):17445
Marie C et al (2018) Oligodendrocyte precursor survival and differentiation requires chromatin remodeling by Chd7 and Chd8. Proc Natl Acad Sci USA 115(35):E8246-e8255
He D et al (2016) Chd7 cooperates with Sox10 and regulates the onset of CNS myelination and remyelination. Nat Neurosci 19(5):678–689
Feng W et al (2017) Chd7 is indispensable for mammalian brain development through activation of a neuronal differentiation programme. Nat Commun 8(1):14758
Jiang X et al (2012) The mutation in Chd7 causes misexpression of Bmp4 and developmental defects in telencephalic midline. Am J Pathol 181(2):626–641
Layman WS et al (2009) Defects in neural stem cell proliferation and olfaction in Chd7 deficient mice indicate a mechanism for hyposmia in human CHARGE syndrome. Hum Mol Genet 18(11):1909–1923
Lin AE, Siebert JR, Graham JM Jr (1990) Central nervous system malformations in the CHARGE association. Am J Med Genet 37(3):304–310
Yu T et al (2013) Deregulated FGF and homeotic gene expression underlies cerebellar vermis hypoplasia in CHARGE syndrome. Elife 2:e01305
Sanlaville D et al (2006) Phenotypic spectrum of CHARGE syndrome in fetuses with CHD7 truncating mutations correlates with expression during human development. J Med Genet 43(3):211–217
Gregory LC et al (2013) Structural pituitary abnormalities associated with CHARGE syndrome. J Clin Endocrinol Metab 98(4):E737–E743
Liu L et al (2014) A novel CHD7 mutation in a Chinese patient with CHARGE syndrome. Meta Gene 2:469–478