Divergent lncRNAs take the lead on pluripotent cell differentiation

A better understanding of the overall transcriptional landscape of human cells has been possible due to the ENCODE and FANTOM projects (1-3). It emerged that mammalian genomes are more intricate than previously suspected and produce a lattice of transcripts, among which only 2% encodes for proteins (4,5). Although alternative splicing and the presence of different Transcription Start Sites can concur to this complexity, a huge part of it can be explained by the existence of short (<30 nt) and long (>200 nt) non-coding RNAs (ncRNAs) which exert their roles without being translated into proteins (6,7). These include the long non-coding RNAs (lncRNAs), which represent a large and diverse class of RNA polymerase II transcripts longer than 200 nucleotides and act as fine-tuners of gene expression by a range of mechanisms (8). LncRNAs have histone-modification profiles, splicing signals, and exon/intron lengths akin to protein-coding genes. However, despite these similarities, lncRNAs are low expressed, preponderant in the nucleus and highly tissue-specific suggesting potential roles in specifying cell identity (9-11) and, when deregulated, in disease (12-14). LncRNAs mode of action can be hinted by their nuclear or cytoplasmic localization due to their unique ability to base-pair, which result in protein coordination and RNA interaction (15). When they are nuclear, lncRNAs can reshape the chromatin by acting in cis (at nearby regions) or in trans (on distant loci). This regulatory potential, in combination with the tissue specificity of lncRNAs, suggests that they can be an active component of a broad epigenetic regulatory network. Indeed, lncRNAs can scaffold distinct histone modification complexes to coordinate discrete functions on specific genomic loci (16-19). Several lines of evidence also suggest that lncRNA-genome interactions can contribute in the organization of three-dimensional nuclear structures (20,21). This is the case of the enhancer RNAs (eRNA), a recent class of lncRNAs which downregulation lead to a reduction in mRNA levels from specific neighbouring genes (22) and a loss of specific enhancer-promoter contacts (23).

Based on their closeness to protein-coding genes, lncRNAs can be classified into (I) sense or antisense, where both overlap another transcript but differ in the selected strand; (II) bidirectional, when its transcript expression starts in close proximity of the neighboring gene; (III) intronic, when their transcription begins from an intron of another transcript; (IV) intergenic, when lncRNA is created between two protein coding genes of an independent transcription unit (24). Although this classification is systematic and limited to the involvement of varied molecular mechanisms in lncRNA expression, it does not provide further information about their modes of action nor for their cellular functions. Evidence suggests that lncRNAs are positively correlated with the expression of the antisense coding genes (9).

Luo and colleagues focused on a particular aspect of lncRNA regulation, which is their ability to act as epigenetic controllers (25). The authors determined the genomic distribution of human and mouse lncRNA genes relative to protein coding loci and focus on antisense head-to-head, also named divergent, lncRNAs (<5 kb to a coding gene). This class of transcripts comprises almost 20% of total lncRNAs. In human, gene ontology analysis revealed that the neighboring protein coding genes associated to divergent lncRNAs were strongly enriched in nuclear functions, as transcription factor activity and sequence-specific DNA binding, as well as embryo development. In line with a possible role in development, the chromatin environment surrounding the divergent lncRNA loci is particularly enriched for chromatin marks that are characteristic of developmental regulators, as the H3K27me3 and the H3K4me3 modifications.

Previous studies have described mammalian lncRNA/mRNA divergent transcription (26,27); however, Luo and colleagues provided novel insight into lncRNA roles at a functional level (25). Indeed, the knock-down of a subset of these lncRNAs in mouse embryonic stem cells (mESCs) led to a downregulation of nearby genes involved in cell fate specification. The authors provide a more detailed analysis on the regulation of the Even-Skipped Homeobox 1 (EVX1) by its divergent partner, Evx1as. Evx1as downregulation by RNAi and antisense oligonucleotides (ASOs), or its genomic deletion by CRISPR/Cas9, led to an attenuation of EVX1 transcription. Conversely, the tethering of Evx1as to EVX1 promoter mediated by a CRISPR system produced a significant increase of the levels of EVX1 transcription, indicating a cis regulatory function of the lncRNA in control of EVX1 gene expression. A deeper investigation of the Evx1as protein interactors revealed that Evx1as facilitates Mediator recruitment and the binding of H3K4me3 and H3K27ac at the promoter, thus shaping an active chromatin state. Single cell analysis during mESC differentiation upon LIF withdrawal revealed that EVX1as is present in most of the cells that activate EVX1 expression. EVX1 is involved in the determination of the character of primitive streak derivatives during gastrulation, in a regulatory network with Goosecoid (GSC) and Brachyury (28). EVX1 and Brachyury stimulate each other’s expression, while EVX1 and GSC are mutually repressive. As a result, EXV1 promotes mesoderm specification at the expenses of endoderm. In their paper, Luo et al. showed that both EVX1 and Evx1as were required for proper mesoderm differentiation. Notably, compared to EVX1, depletion of Evx1as seemed to produce a stronger negative effect on the expression of mesoderm genes, leaving an open possibility of a broader function for this divergent lncRNA.

In conclusion, in addition to cis regulatory DNA elements as promoters or enhancers, divergent lncRNAs may provide another layer of transcription regulation. Fine-tuning of gene expression is extremely important during early phases of embryonic development, when key factors must be activated, maintained and eventually repressed in a tightly regulated manner, in space and time. Increasing evidence suggests that the contribution of the non-coding portion of the transcriptome to this fine-tuning is considerable. Certainly, lncRNAs represent flexible, mobile, and transient molecules thus providing a convenient mean to precisely regulate nearby gene expression in a site-specific way. We believe that more examples of lncRNAs with a key function in development and differentiation, such as those described in the paper by Luo and colleagues, will come in the next future.