"Seq-ing" insights into the epigenetics of neuronal gene regulation

Abstract

The epigenetic control of neuronal gene expression patterns has emerged as an underlying regulatory mechanism for neuronal function, identity, and plasticity, in which short- to long-lasting adaptation is required to dynamically respond and process external stimuli. To achieve a comprehensive understanding of the physiology and pathology of the brain, it becomes essential to understand the mechanisms that regulate the epigenome and transcriptome in neurons. Here, we review recent advances in the study of regulated neuronal gene expression, which are dramatically expanding as a result of the development of new and powerful contemporary methodologies, based on next-generation sequencing. This flood of new information has already transformed our understanding of many biological processes and is now driving discoveries elucidating the molecular mechanisms of brain function in cognition, behavior, and disease and may also inform the study of neuronal identity, diversity, and neuronal reprogramming.

Figure 1. A high diversity of next-generation or deep sequencing approaches is currently available for profiling genomes, epigenomes, methylomes, and transcriptomes

A plethora of deep sequencing approaches are now available, ranging from approaches to map the primary sequence of DNA (whole-genome-seq and exome-seq), mapping DNA methylation marks (meDIP-seq, 5-hmC-seq, and many others), profiling chromatin structure (MNase-seq, DNaseI-seq, and FAIRE-seq), profiling all the different stages of the transcriptome (GRO-seq, RNA-seq, and ribo-seq), profiling transcription factors, cofactors, and histone marks (ChIP-seq), profiling RNA interactions to the genome or the transcriptome (ChIRP-seq and CLIP-seq, and variants), to finally profile the structure of the genome in the tridimensional space (ChIA-PET, HiC, and several others). All these approaches are now available for the neurobiology community and are primed to revolutionize the field.

Figure 2. Schematic model of regulation of gene expression by REST/NRSF

In embryonic stem cells, REST/NRSF is associated with RE1-containing sequences and assures silencing of target genes by tethering repressive components, including HDAC1/2, LSD1, G9a, Suv39h1, CtBP, MeCP2 and Brg1 among others. During differentiation towards neuronal cell fate, the REST protein level is dramatically reduced via several mechanisms such as direct proteosomal degradation of the protein as well as via transcriptional repression and miRNA-dependent degradation of the mRNA. Remaining low levels of REST/NRSF protein are excluded from nucleus through interaction with Huntingtin.

Figure 3. A. Dynamics of DNA methylation

DNA methylation changes can be brought about by diverse signals such as neuronal activity or during development. Active promoters are generally unmethylated (open circles) allowing the binding of transcription factors (TF), recruiting RNA polymerase (RNA Pol II) and other factors for transcription to occur. Upon methylation (closed black circles), Methyl binding proteins (MBPs) are recruited to promoters and recruit repressive machinery such as histone deacetylases (HDACs) and Corepressors (CoRep), which lead to reduced transcription. In some cases, further recruitment of other enzymes such as the H3K9HMT Suvar39h (K9HMTs), which deposit the H3K9me3 mark on histone tails, can lead to further repression by recruiting HP1, condensation of the chromatin, and spreading of the repressive state. Similarly, DNA methylation changes can occur on enhancers, which when unmethylated (open circles), allow the binding of transcription factors (TF) and other proteins required for enhancer activity. B. DNA methylation variants. A series of enzymes are capable of demethylating 5-methyl Cytosines (5-mC) to an unmethylated state, with various intermediates. TET1, a member of the Tet family of proteins, is a 5-mC dioxygenase responsible for catalyzing the conversion of 5-mC to 5-hydroxymethylcytosine (5-hmC) and further to 5-formylcytosine (5-fC) and/or 5-carboxylcytosine (5-caC). Alternatively, 5-hmC can be deaminated to 5-hydroxymethyluracil (5-hmU). These derivatives of 5-hmC can be further converted to an unmethylated state via various mechanisms such as glycosylation by TDG and the base excision repair machinery (BER). These various modifications highlight the newly found diversity and dynamics in the DNA methylation landscape; however, the precise roles of these modifications have yet to be determined. The suggested role of these modifications is described below each of them.

Figure 4. ncRNAs function in the brain

A. Schematic illustration of various classes of RNA species derived by different genomic locations, including coding, short and long non-coding RNAs. Often lncRNAs are defined based on their location relative to the coding gene as divergent, antisense, intronic, and bidirectional. Intergenic lncRNAs are transcribed by separate transcriptional units. B. lncRNAs mechanisms of action: “guide” histone modifying enzymes to chromatin (XIST); scaffold molecules that bring together proteins complexes with different enzymatic activities (HOTAIR); “looping” of distant genomic regions through the recruitment of protein complexes (enhancer-like RNAs); inhibition of transcription factors binding to their cognate DNA motifs (GAS5); transcriptional activation by interacting with TFs; nucleation of nuclear structures (MALAT1) or “guide” of specific histone mark readers in distinct nuclear structures (TUG1). C. Model of RA-induced transcription of riRNAs in stem cells, where they target various mRNAs in an AGO3-dependent mechanism. D. Many lncRNAs have been linked to various neurological disorders, but their mechanism of action remains elusive.

Figure 5. The epigenome to assess neuronal identity of reprogrammed cells

A. There are three common strategies to generate a neuron in vitro (differentiated from ESCs, iPSCs, or directly transdifferentiated from somatic cells), and each might be accomplished by different protocol variants (represented as 1-4). However, it remains unclear how similar is the level of conversion, efficiency, and reproducibility to the neural lineage between them, and which is the most similar to endogenous neurons (center of the image). Since transcriptomes may miss some identity features that epigenomes reveal, we propose a systematic use of the second as potential “barcodes” to establish the most appropriated protocol. B. Neurons show an especially high identity diversity or heterogeneity, a systematic effort of epigenetic profiling might be a useful tool to catalog and distinguish between these different identities. Furthermore, such catalogs may facilitate the search and identification of the best protocols to generate these different identities in vitro (bottom).