Emerging Themes in Epigenetics: Implications for Understanding and Treating Nervous System Disorders

The central aim of neurobiology is to characterize the human CNS, a biological machine of unparalleled sophistication, which is comprised of approximately 1.7 × 10¹¹ neural cells, including neural stem and progenitor cells as well as mature neuronal and glial subtypes. These highly specialized cell types form a diverse array of functional regions that include, for example, stem-cell niches and neocortical layers, and are further organized into local and more widely distributed neural networks, in which, an average neuron may have between 5000 and 200,000 synapses. Moreover, the CNS exhibits exquisite degrees of sensitivity to interoceptive and environmental signals that can induce homeostatic and plasticity responses at molecular (e.g., gene expression), cellular (e.g., synaptic strength) and more hierarchical (e.g., neural network activity and topology) levels. These features imbue the CNS with an unmatched ability to process salient informational cues and a unique functional repertoire.

The advent of epigenomics has offered insights towards the understanding of CNS structure and function, including how epigenetic factors are responsible for generating neural cell identity, promoting neural network connectivity and plasticity, and performing higher-order functions, including learning and memory. It has also begun to elucidate the roles of epigenetic mechanisms in the transgenerational inheritance of complex cognitive and behavioral traits and disease susceptibility [1].

The transformation in our understanding of RNA biology and its roles within the CNS in particular, is one of the most exciting developments of the epigenomic era. Indeed, noncoding RNA (ncRNA) dynamics are likely to represent the principal engine responsible for driving human CNS processes. Several lines of evidence support this hypothesis. Because the proportion of noncoding sequences rises as a function of developmental complexity comprising more than 98% of the human genome, most of which is transcribed into ncRNAs that are expressed within the CNS, it is likely that these ncRNAs provided the substrates for evolutionary innovations in the brain. In fact, 47 of 49 regions in the human genome exhibiting accelerated changes in the human lineage are noncoding. Intriguingly, HAR1A is a ncRNA transcribed from one of these regions that is specifically co-expressed in Cajal–Retzius cells with the critical neural developmental factor reelin, which mediates neuronal migration and positioning in the developing neocortex and is implicated in the pathogenesis of many neurological and psychiatric diseases [2]. Large-scale efforts including the Encyclopedia of DNA Elements (ENCODE) project [3] and the Functional Annotation of the Mammalian Genome (FANTOM) consortium [4,5] have yielded an increasingly complex view of genomic architecture and ncRNAs. In addition to the relatively small number of protein-coding genes, the genome encodes a vast array of interleaved ncRNAs. The precise number is unknown; however, reasonable estimates suggest, astonishingly, that there are hundreds of thousands if not a million or more.

Many of these ncRNAs are expressed in the CNS in a developmentally regulated, cell type- and subcellular compartment-specific, and activity-dependent manner. These include many classes of short ncRNAs (e.g., microRNAs and small nucleolar RNAs), whose roles in mediating critical biological processes have been the focus of intense scrutiny. For example, microRNAs bind most commonly to complementary sequences in the 3´-UTR of target mRNAs, repressing translation of these transcripts or sequestering them for storage or degradation. Intriguingly, mRNAs expressed in the brain have particularly long 3´-UTRs [6], and critical neural genes, including brain-derived neurotrophic factor, produce mRNAs with short and long 3´-UTRs that are differentially localized to somata and dendrites, highlighting the complexity and context specificity of 3´-UTR-based regulation in the CNS [7].

Other classes of short ncRNAs have also been described (e.g., Piwi-interacting RNAs [piRNAs], promoter-associated small RNAs, small RNAs derived from small nucleolar RNAs and transcription initiation RNAs) [8]. The roles of these ncRNAs are not well characterized; however, it is likely that most, if not all, are involved in epigenetic regulation in the CNS [9]. For example, piRNAs were initially thought to silence mobile genetic elements in the germline; however, piRNAs and their interactors (mammalian PIWI proteins) have now been found in certain somatic cells [10]. It is possible that piRNAs may also be present in neuronal cells, where they may participate in regulating mobile genetic elements, including L1 retrotransposons, which are implicated in generating neuronal cell diversity [9,11,12]. In fact, C15orf2 is a gene harboring several piRNAs, that exhibits a signature of positive selection in humans, is expressed in the fetal brain and may play a role in Prader–Willi syndrome, a neurodevelopmental disorder [13,14].

In contrast to short ncRNAs, long ncRNAs (lncRNAs; >200 nucleotides), such as HAR1A, have received far less attention despite the presence of hundreds of thousands of these lncRNAs in the genome, a large percentage of which are expressed in the CNS. Because of their length, these factors have many more dimensions in orthogonal conformational (analog) and sequence (digital) spaces and are, therefore, capable of much more complex molecular functions than short ncRNAs [15]. lncRNAs seem to play an extremely broad array of structural, regulatory and catalytic roles, including, for example, the formation of nuclear subdomains during neural cellular differentiation (e.g., Neat1 and Gomafu), the modulation of local protein synthesis in postsynaptic dendritic compartments (e.g., BC1/BC200), nuclear–cytoplasmic shuttling (e.g., BACE1-AS and NRON) and X-chromosome inactivation (e.g., Xist). Moreover, the genomic context from which certain lncRNAs are transcribed seems to be important for mediating their function. Some lncRNAs are found in intergenic regions while others are derived from genomic loci encompassing protein-coding genes, including many encoding key neural factors. An increasing number of these lncRNAs are thought to have roles in chromatin remodeling, transcriptional regulation and post-transcriptional processing at local and long-distance genomic sites through both cis- and trans-acting mechanisms. Furthermore, many lncRNAs are bound to chromatin-modifying complexes, including PRC2 and CoREST-containing complexes, suggesting that lncRNAs may act as molecular beacons or guideposts for genomic site-specific epigenetic remodeling. Interestingly, both PRC2 and CoREST play roles in mediating neural cell-fate decisions [16–19], highlighting the intimate relationship between ncRNAs and the establishment of neural cell identity and functional diversity [20].

In addition, RNA editing seems to be particularly important in the human CNS, where it allows for the environmentally responsive diversification of RNAs. RNA editing occurs not only in mRNAs that are associated with neurotransmission and synaptic plasticity, as was initially believed, but predominately in ncRNAs [21]. Interestingly, the amount of RNA editing in humans is significantly greater than in mice and takes place largely in transcripts derived from Alu sequences and L1 elements, which are responsible, in part, for the expansion of noncoding sequences in the human genome. Intriguingly, DNA-editing enzymes may be involved in modulating these mobile genetic elements, and DNA-editing enzymes have significantly expanded during primate evolution, raising the possibility that coupled RNA and DNA editing represent mechanisms for recoding productive environmentally mediated changes back into the genomes of neural cells, including post-mitotic neurons and, perhaps, for promoting higher cognitive functions such as learning and memory [22]. Post-transcriptional RNA dynamics including RNA modifications, axodendritic transport and local mRNA translation and associated synaptic plasticity events are coordinated by higher-order control systems that utilize RNA motifs and RNA binding proteins (RBPs) to form ribonucleoprotein complexes, which can synchronize the distribution and function of mRNAs, ncRNAs, proteins and related factors [23]. These RNA operons and higher-order regulons are implicated in mediating bidirectional axodendritic transport that may be responsible for relaying productive RNA editing events from the synapse back to the nucleus for DNA recoding. RNAs may also be involved in intercellular communication through the active transport of RNAs to adjacent nerve cells, to more distant sites and possibly even to the germline, thus promoting neural network synchronization, somatic signaling and feedback regulation and multigenerational inheritance [24].

It has been suggested that these complex RNA dynamics may have fueled evolutionary innovations in the human CNS [1]. Indeed, the brain is a conspicuous consumer of energy, and the bioenergetic cost of RNA is less than DNA or protein. Also, RNA serves as a highly flexible, sensitive, high fidelity information encoding and functional molecule that can rapidly be transcribed, modified, stored, transported and degraded, processes that can be regulated by many cascades of feedback and feed-forward loops in response to interoceptive and environmental stimuli. This dynamic matrix of possible interactions probably mediates the robustness and plasticity that characterizes CNS function [15,25].

Considering these observations, it is not surprising that perturbations in RNA metabolism and function are increasingly being recognized as the primary mechanisms responsible for causing a number of CNS diseases. These pathogenic processes include but are not limited to, often subtle, disruptions in the pathways mediating mRNA and ncRNA transcription, RNA editing, splicing and transport, ribonucleoprotein formation, mRNA translation and RNA quality control [26,27].

For example, mutations in genes encoding the RBPs TDP-43 and TLS/FUS are responsible for causing amyotrophic lateral sclerosis and frontotemporal lobar dementia. The specific roles played by RBPs in the pathophysiology of these disorders are not well characterized, however, the functions of TLS/FUS suggest intriguing epigenetic links between neurodegeneration, DNA damage and cell cycle regulation. In fact, TLS/FUS can be recruited to the promoter of cyclin D1 by a lncRNA that is induced in response to DNA damage, where it represses the transcription of this cell cycle regulatory factor [28]. Moreover, TLS/FUS is also likely to play a role in other CNS disorders because it is responsible for regulating RNA expression and function through more global modulation of the spliceosome and of RNA polymerase II and III activities, which mediate the transcription of mRNA and ncRNA species, respectively [29]. In fact, TLS/FUS constitutes a major component of the abnormal neuronal intranuclear inclusions found in neurodegenerative disease associated with microsatellite repeat expansions including spinal cerebellar ataxia types 1–3, dentatorubral–pallidoluysian atrophy and Huntington‘s disease [30,31].

These unstable expansions of tandem repeats occur in protein-coding, as well as in noncoding regions, and cause at least 22 distinct neurological diseases. Mutant mRNAs and ncRNAs derived from these sequences may have pathological effects that include the disruption of interactions between these RNAs and trans-acting factors, such as RBPs. This RNA dominant toxicity is now believed to be one of the principal mechanisms mediating the pathogenesis of this spectrum of neurodevelopmental, neurodegenerative and neuromuscular disorders. Interestingly, different degrees of repeat expansion within the Fmr1 gene locus are responsible for causing neurodevelopmental (i.e., fragile X syndrome) or neurodegenerative (i.e., fragile X-associated tremor/ataxia syndrome) disorders. FMRP is an RBP transcribed from this gene locus that is critical for mediating synaptogenesis and neuronal plasticity through diverse effects on RNA dynamics including transport, translational control and miRNA biogenesis and function. In addition, various lncRNAs (i.e., FMR4 and ASFMR1) are also transcribed from the Fmr1 gene locus, and these lncRNAs may be independently associated with fragile X syndrome and fragile X-associated tremor/ataxia syndrome [1].

Of note, mitigation of RNA toxicity has been demonstrated in proof-of-concept studies of various therapeutic approaches aimed at treating these disorders. These strategies include, for example, accelerating the turnover of mutant RNAs, overexpressing levels of trans-acting factors in order to overcome the effects of toxic RNAs that may sequester these factors and inhibiting the molecular interactions between RNA expanded repeats and their binding partners [32].

Together, these observations highlight the complexity of RNA-based mechanisms operating within the CNS and imply that better characterizing these processes is critical for understanding the pathogenesis of diverse neurological and psychiatric disorders. Furthermore, they support the vigorous advancement of efforts aimed at developing novel therapeutic strategies for selectively modulating RNA dynamics including, for example, their expression profiles, editing patterns, axodendritic transport, intercellular trafficking, site-specific genomic targeting of chromatin remodeling complexes and specific biological functions and also for designing and delivering RNA-based agents that exploit the unique biophysical properties of these multifunctional molecules, such as their sequence-specific (digital) and conformational (analog) features.

Financial & competing interests disclosure

Mark F Mehler is supported by grants from the National Institutes of Health (NS38902 and MH66290), as well as by the Roslyn and Leslie Goldstein, the Mildred and Bernard H Kayden, the FM Kirby and the Alpern Family Foundations. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.

Additional information

Funding

Mark F Mehler is supported by grants from the National Institutes of Health (NS38902 and MH66290), as well as by the Roslyn and Leslie Goldstein, the Mildred and Bernard H Kayden, the FM Kirby and the Alpern Family Foundations. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed. No writing assistance was utilized in the production of this manuscript.