Mammalian evolution exhibits extraordinary acceleration in phenotypic complexity while displaying genetic conservation. Indeed, protein-coding genes are 90% identical between mice and human, and the question arises what might be the mechanisms by which phenotypic complexity increases while the number of proteins does not? An interesting possibility is that the increase in regulatory complexity, as reflected in a more sophisticated spatio-temporal regulation on gene expression, could generate more diverse combinations of proteins and a greater number of distinct phenotypes. Importantly, unlike protein coding genes, long noncoding RNAs (lncRNA) exhibit only 50% similarity between mice and human.1 Are these RNA species involved in such regulation?
In recent years, lncRNA have captured the attention of biologists. LncRNA are highly abundant in the human genome, and it has been demonstrated that they have fundamental roles in a variety of biological processes. In particular, lncRNA have been shown to participate in gene regulation by binding to proteins, such as chromatin regulators, and targeting them to particular regions in the genome.2,3 The interaction between lncRNA and specific proteins often depends on an explicit sequence within the lncRNA,4,
Two recent works11,12 have demonstrated that a number of lncRNAs act in cis to regulate the transcription of neighboring genes. By analyzing 12 lncRNA loci in mouse embryonic stem cells, Engreitz et al.11 found that knocking out the promoter of five of these lncRNAs cis-modulated the expression of nearby genes, which were located within 5–71 kb from the promoters. To assess whether these cis effects were driven by the lncRNAs sequence, by the transcription process itself, or by DNA regulatory elements present in the promoter, polyadenylation signals were inserted in the studied lncRNA genes to prevent their transcription while preserving the promoter integrity. In a number of cases, the insertion of polyadenylation signals did not impact the transcription of neighboring genes,11 arguing that in these cases transcription of lncRNAs is not involved in modulating nearby gene expression. These experiments thus support the notion that enhancer-like DNA regulatory elements are present in such lncRNA promoters. In other cases, termination of lncRNA transcription impacted neighboring gene transcription in cis,11,12 indicating that the transcription of some lncRNAs is required for the observed cis effect. However, in all cases, specific lncRNA sequences were not required to mediate such effects, as exon deletion or sequence replacement did not affect nearby gene expression.11,12 Moreover, while a number of lncRNAs are not conserved, the DNA regulatory elements with enhancer activity present in lncRNA promoters are highly conserved,11 highlighting the functional importance of these cis-regulatory elements.
Such cis gene regulatory mechanism is physiologically relevant as shown in mouse models.12 Anderson et al.12 demonstrated in vivo that the transcription of the heart development regulator Hand2 is controlled in cis by the lncRNA upperhand (Uph), which is located 150 bp distant on the opposite strand. Gene-targeted knockout of Uph, but not knockin of a heterologous DNA sequence into the Uph locus, blocked expression of Hand2 in mouse heart and gave rise to a similar cardiac phenotype as Hand2 knockout embryos. The cis effect is therefore dependent on Uph transcription but not on its sequence. Uph transcription was found to be required to maintain histone marks characteristic of super-enhancers, and to recruit the GATA4 transcription factor to the Hand2 cardiac enhancer.12
Together, these works underscore the functional importance of lncRNA genes in fine tuning the expression of an adjacent protein-coding gene in a sequence-independent manner. The idea of sequence-independent functions of lncRNA is supported by a recent study, which showed that there is no evidence for conservation of the secondary structure of known functional lncRNA.13 Assuming that RNA–protein interaction is structure dependent, it is tempting to speculate that lack of structural conservation in lncRNA indicates that their major functions, from an evolutionary perspective, are related to their cis functions. In accordance, Kapusta and Feschotte14 proposed an evolutionary hierarchy by which lncRNA acquire functions; first. they function by regulating an adjacent gene in cis, and subsequently, they acquire trans functions (Figure 1). This is in line with higher sequence conservation observed in older mammalian-conserved lncRNA as compared with younger, humanoid-specific lncRNA.15
Analysis of human lncRNA sequences indicate that they have originated from transposable elements (TE) as 75% of human lncRNA have traces of a TE.16 It is therefore conceivable that, during evolution, insertion of a TE at a specific location could affect the expression of its adjacent genes, their promoters, and the local chromatin landscape. In cases where this insertion increased fitness, the TE element would be selected for. Indeed, regulation of the chromatin landscape occurs at several levels as well. For example, the ‘ripple effect’ shows that the expression of a particular gene can affect the expression of its neighbors up to 100 kb away.17 Together, with the fact that neighboring genes are evolutionarily linked,18 we argue that the genomic location of lncRNA may contribute significantly to their function and their birth.
Properties of lncRNA have important implications to how we study these genes. In the study of protein coding genes, the gene sequence provides the sequence of the protein product, the domains it harbors, its conservation, and rate of changes through evolution, all of which help elucidate gene function. However, in the case of lncRNA, it would seem that there is less functional information embedded in the sequence of the gene, and their major functions might be related to their genomic location. Therefore, one of the major challenges in the field of lncRNA is to be able to integrate the genomic and transcriptomic data to infer possible functions for lncRNA genes in light of the possibility that they might be noncoding and not coded.
We wish to thank Shimon Bernstein for insightful discussion. This study was funded by research grants from the Israeli Science Foundation (ISF 1124/15) to BR.