To understand the roles of 5hmC and TETs in regulation of neural gene expression, several studies have mapped the genomic distribution of 5hmC in various neural cell types and tissues over the course of embryonic and postnatal development (Jin et al., 2011; Szulwach et al., 2011; Khare et al., 2012; Hahn et al., 2013; Lister et al., 2013). In the embryonic mouse cortex, 5hmC levels increase as neural progenitor cells develop into mature neurons (Hahn et al., 2013). Interestingly, this increase is not necessarily accompanied by an increase in unmodified cytosine levels, suggesting that 5hmC can be a stable epigenetic mark in neurons and not merely a DNA demethylation intermediate (Hahn et al., 2013). Consistent with the notion that 5hmC is derived from 5mC, genomic regions in the fetal mouse brain that are enriched for 5hmC also tend to be enriched for 5mC. Notably, many of these regions become depleted of both marks in the adult mouse (Lister et al., 2013), demonstrating that 5hmC facilitates DNA demethylation in the developing brain.

5hmC levels increase in various mouse and human brain tissues over the course of life (Szulwach et al., 2011), and may have implications for neurodegenerative diseases. 5hmC is enriched in gene bodies and promoters, depleted from intergenic regions and transcription start sites, and is deposited at brain-specific enhancers (Jin et al., 2011; Szulwach et al., 2011; Hahn et al., 2013; Lister et al., 2013; Cui et al., 2020). The presence of 5hmC in gene bodies is associated with increased gene expression, suggesting that TET enzymes and 5hmC contribute to a transcriptionally permissive state of chromatin in the brain (Jin et al., 2011; Szulwach et al., 2011; Hahn et al., 2013; Lister et al., 2013). 5hmC also demarcates intron-exon boundaries in human brain cells and marks constitutively expressed exons, suggesting a potential role in control of splicing (Khare et al., 2012). Genes with high levels of 5hmC, for example Syt1 and Nav2, belong to functional categories critical for nervous system function, such as synaptic transmission and neurogenesis (Khare et al., 2012; Hahn et al., 2013). 5hmC is also associated with repetitive elements as it is enriched at SINE and LTR elements in the cerebellum and hippocampus, and depleted from LINE elements in the cerebellum (Szulwach et al., 2011). Enrichment at SINE and LTRs increases over postnatal life in the cerebellum (Szulwach et al., 2011), indicating a possible role in regulation of repetitive element activity in the brain. Indeed, Tet2/3 knockdown reverses loss-of-Uhrf1-mediated increased DNA hydroxymethylation and activation of IAP elements in NPCs (Ramesh et al., 2016). Together, these observations support important roles for 5hmC and TETs in mammalian neurogenesis and brain function.

Studies of embryonic stem cell (ESC) differentiation have suggested a critical role for TET enzymes and 5hmC in neural lineage commitment. Deficiency of all three TETs in ESCs compromises pluripotency and Tet1/2/3 triple knockout (TKO) ESCs fail to form neural pigmented epithelium in teratoma assays, though they are able to form other neural tissue types (Dawlaty et al., 2014). These cells fail to contribute to nervous system structures when injected into wild type blastocysts to form chimeras (Dawlaty et al., 2014). Consistently, Tet TKO mouse embryos and ESCs differentiated toward the neural lineage have reduced neuroectodermal and increased mesodermal gene expression, in part due to failure to inhibit Wnt signaling (Li et al., 2016). Likewise, TET TKO human ESCs exhibit aberrant neuroectodermal gene expression when differentiated toward the neural lineage and fail to demethylate the PAX6 promoter, a transcription factor critical for neurodevelopment (Verma et al., 2018). These studies support a requirement for TET enzymes in the commitment of ESCs to a neural fate, an idea further supported by studies of TET genes in ESC specification to neural progenitor cells (NPCs). TET enzymes, in particular TET2, regulate enhancer methylation during differentiation of ESCs to NPCs (Hon et al., 2014). Though Tet2 knockout ESCs can successfully differentiate into NPCs, these cells exhibit delayed induction of neural gene expression programs accompanied by enhancer hypermethylation and reduced histone H3 lysine 27 acetylation (Hon et al., 2014). This is in line with DNA hypermethylation in the embryonic cerebral cortex of Tet2 knockout mice (Lister et al., 2013). TET3 plays a role in the epigenetic regulation of NPC specification and maintenance of cellular identity (Montibus et al., 2020; Santiago et al., 2020). During differentiation of mouse ESCs to NPCs, the catalytic activity of TET3 promotes expression of histone demethylase Kdm6b, an epigenetic regulator critical for gene regulation during neurogenesis (Montibus et al., 2020), and loss of TET3 promotes NPC apoptosis (Li T. et al., 2015). Knockdown of Tet3 in NPCs promotes de-repression of pluripotency genes Oct4 and Nanog, implicating TET3 in the maintenance of NPC identity (Santiago et al., 2020). These studies implicate TETs in the epigenetic regulation of NPC biology.

In adult NPCs, different TET paralogs have unique functions, highlighting some non-redundant and context-specific roles. Loss of TET2 increases the proliferation of adult NPCs and reduces their differentiation into neurons and astrocytes in vivo, indicating that TET2 promotes NPC differentiation (Li et al., 2017). Deletion of Tet3 decreases NPC proliferation in the subventricular zone of the mouse cortex and promotes astrocytic differentiation, consistent with a role for TET3 in maintaining NPC identity (Montalban-Loro et al., 2019). Tet1 knockout mice have fewer NPCs in the dentate gyrus of the hippocampus and conditional deletion of Tet1 or Tet2 in NPCs compromises hippocampal neurogenesis (Zhang et al., 2013; Gontier et al., 2018). While most functions of TETs in NPCs are attributed to their enzymatic activity (Zhang et al., 2013; Li et al., 2017; Montibus et al., 2020), some functions are independent of enzymatic activity, such as transcriptional repression of the imprinted gene Snrnp by TET3 (Montalban-Loro et al., 2019). Investigating these dual roles of TETs and dissecting their requirements in NPC biology and neurogenesis will be essential.

Although global or neural-specific loss of each Tet gene in mice influences NPC biology, it does not block neurogenesis or cause gross neuroanatomical defects (Rudenko et al., 2013; Zhang et al., 2013). However, combined loss of TET1/2 and TET1/3 causes exencephaly and holoprosencephaly in some embryos, respectively (Dawlaty et al., 2013; Kang et al., 2015) suggesting redundancy between TETs in neurogenesis that warrants further investigation. Findings from other organisms have also supported a role for TET enzymes in embryonic neurogenesis. Xenopus laevis embryos depleted of TET3 are microcephalic and eyeless with deregulation of neurodevelopmental programs leading to aberrant expression of neuronal, eye, neural crest, and sonic hedgehog signaling genes (Xu et al., 2012). Moreover, tet2/3 mutant zebrafish exhibit abnormal brain and eye morphology (Li C. et al., 2015) and impaied retinal neurogenesis, partly due to overactive Notch and Wnt signaling (Seritrakul and Gross, 2017). Aberrant expression of mesodermal genes was also observed in tet2/3 mutant retinas (Seritrakul and Gross, 2017), a finding similar to those in Tet1/2/3 knockout embryos (Li et al., 2016). Importantly, TET enzymes mediate demethylation of conserved developmental enhancers in brain during the phylotypic stage of vertebrates, as demonstrated in zebrafish, Xenopus tropicalis, and mouse (Bogdanovic et al., 2016). Together, these findings support highly conserved and overlapping functions for TETs in neurodevelopment.

In addition to roles in regulation of NPC biology, TETs and 5hmC are important in postnatal neurodevelopment and mature neurons. As previously mentioned, 5hmC accumulates over the course of life (Szulwach et al., 2011). During development of mouse olfactory bulb neurons, which occurs throughout life, 5hmC is enriched in neurons relative to immature cells and is associated with increased neurodevelopmental gene expression (Colquitt et al., 2013). Likewise, 5hmC increases over the course of postnatal retinal maturation, and is enriched at neurogenesis genes (Perera et al., 2015). In the cerebellum, 5hmC increases during an important period of neuronal circuit formation, and TET1 and TET3 are required for proper branching of granule cell dendrites (Zhu et al., 2016). Chimeric Tet3 knockout mice generated by injection of Tet3 sgRNAs in one cell of a two-cell-stage embryo develop histologically normal cerebral cortices composed of Tet3 wild type and knockout cells but exhibit abnormal electrophysiology in recordings of excitatory and inhibitory neurotransmission, suggesting that TET3 is required for developmental synapse and circuit formation (Wang et al., 2017). These studies implicate TETs in shaping the epigenetic landscape during specification of mature neural cell types after birth and in the development of higher order structures, including neuronal circuits.

5hmC and TET enzymes have also been shown to be highly dynamic within post-mitotic neurons. Cortical 5hmC has cell-type specific distributions associated with differential gene expression (Kozlenkov et al., 2018), and the ability of TETs to promote active DNA demethylation and alter gene expression in response to neuronal activity and to influence behavior has been the subject of extensive study (Guo et al., 2011; Kaas et al., 2013; Rudenko et al., 2013; Zhang et al., 2013; Li et al., 2014; Yu et al., 2015). Tet1 expression is downregulated in response to neuronal activity in hippocampus where it regulates spatial memory and fear memory extinction (Guo et al., 2011; Kaas et al., 2013; Rudenko et al., 2013; Zhang et al., 2013). Hippocampal neurons upregulate Tet3 to initiate active DNA demethylation in response to neuronal stimulation (Yu et al., 2015). In cortical neurons, Tet3 is upregulated during fear extinction learning and, like Tet1 in the hippocampus, is required for fear memory extinction (Li et al., 2014). Fear extinction learning is accompanied by Tet3-mediated upregulation of the Gephyrin gene and a transcriptionally-permissive reshaping of chromatin around this locus (Li et al., 2014). Loss of Tet3 is sufficient to produce anxiety-like behaviors in mice, partly due to increased expression of immediate early genes like Npas4 (Antunes et al., 2020), a role that is opposite to the anxiolytic and anti-depressant effects of Tet1 (Feng et al., 2017). In general, the mechanistic basis by which TET enzymes influence behavior is, in part, due to reshaping of neuronal 5mC and 5hmC landscapes in response to activity. This remodeling of the epigenome is required for proper expression of genes involved in memory consolidation and synaptic function, such as Bdnf and Arc, and is sufficient to alter the electrophysiological properties of neurons (Guo et al., 2011; Kaas et al., 2013; Rudenko et al., 2013; Zhang et al., 2013; Li et al., 2014; Yu et al., 2015). In post-mitotic cerebellar neurons, 5hmC dynamics can influence recruitment of key gene regulatory factors. For example, 5hmC in gene bodies is associated with reduced MeCP2 occupancy and increased gene expression, possibly due to loss of MeCP2 repression (Mellen et al., 2017). Of note, reduced MeCP2 occupancy is specifically associated with 5hmC at gene body CpG dinucleotides, whereas 5hmC at CpA sites flanking enhancers retains MeCP2 binding. This highlights the ability of 5hmC to influence recruitment of gene regulatory factors in a sequence-dependent manner (Mellen et al., 2017). Moreover, findings that common MECP2 mutations in Rett syndrome disrupt MeCP2 binding to 5hmC has implications for this mark in Rett syndrome pathogenesis (Mellen et al., 2012; Brown et al., 2016). Together, these studies propose crucial roles for 5hmC and TET enzymes in mature neuronal function.

Compelling evidence for the importance of TETs in neurodevelopment and brain function is the identification of TET gene mutations and alterations in 5hmC levels in human neurodevelopmental and neurodegenerative disorders. Mutations in TET3 were recently identified to underlie an inherited syndrome of intellectual disability and craniofacial abnormalities (Beck et al., 2020). While most mutations are in the catalytic domain and are sufficient to impair enzymatic activity, some are outside of this domain, and one mutation does not affect catalytic activity (Beck et al., 2020), underscoring the importance of TET3 catalytic and non-catalytic functions in human neurodevelopment. Interestingly, the clinical characteristics of patients with TET3 deficiency resemble those of patients with Tatton-Brown-Rahman syndrome and Sotos syndrome, caused by mutations in DNMT3A and NSD1, respectively (Kurotaki et al., 2002; Tatton-Brown et al., 2014). This highlights the general importance of DNA and histone methylation dynamics in human craniofacial and neural development. Other TET mutations have been observed in individuals with intellectual disability. Mutations in TET1 were identified in consanguineous Pakistani and Iranian families with familial intellectual disability syndromes (Harripaul et al., 2018), and a germline TET2 variant in an individual diagnosed with intellectual disability and delayed verbal comprehension in the absence of any other known genetic causes (Kaasinen et al., 2019). Together, these findings support an important role for TETs in the etiology of neurodevelopmental disorders and intellectual disability.

In addition TET enzymes and 5hmC are recurrently dysregulated in neurodegenerative conditions and in aging brain. Induced pluripotent stem cell-derived NPCs and neurons from Alzheimer’s disease (AD) patients exhibit differential hydroxymethylation at genes associated with neurodevelopment and synaptic function, including at known AD susceptibility loci, compared to cells derived from healthy controls (Fetahu et al., 2019). Consistently, presumptive loss-of-function mutations in TET2 have been identified in patients with early onset AD and frontotemporal dementia (Cochran et al., 2020). Interestingly, TET2 promotes proinflammatory gene expression in microglia and TET2 expression is increased in microglia associated with amyloid beta plaques in the brains of AD patients and mouse models (Carrillo-Jimenez et al., 2019). Thus, the positive and negative roles of TET2 in AD are likely specific to distinct stages in clinical course and cell types. TET variants or dysregulation have also been implicated in Parkinson’s disease (PD). TET1 mutations were reported in a Chinese cohort of PD patients (Shu et al., 2019). Intriguingly, increased expression of TET2 and increased 5hmC at neural enhancers is observed in prefrontal cortex of patients with PD and Tet2 knockout mice are protected from inflammatory damage to the substantia nigra (Marshall et al., 2020). Conversely, Tet2 expression declines in the hippocampus of aging mice and is associated with age-related cognitive decline (Gontier et al., 2018). Remarkably, restoration of hippocampal Tet2 expression by stereotactic lentivirus injection is sufficient to improve cognitive function in aged mice (Gontier et al., 2018). Moreover, recent findings that TET1 and TET2 are required for axonal regeneration by reprogramming factor expression highlights their potential as therapeutic targets (Lu et al., 2020). These observations in human disease and mouse models warrant further studies to clarify the discordant roles of TET enzymes in the etiology of neurodegenerative diseases and aging.

In addition to their roles in neurodevelopmental and neurodegenerative disorders, TET enzymes are associated with addictive behaviors in humans. TET1 expression is decreased in the nucleus accumbens (NAc) of humans suffering from cocaine addiction and cocaine administration to mice is sufficient to alter 5hmC at enhancers in NAc (Feng et al., 2015). Tet expression in NAc is also responsive to methamphetamine administration in rats (Jayanthi et al., 2018). Further work is necessary to clarify the role of TETs in mediating addictive behaviors.