Long non-coding RNA and chromatin remodeling

Abstract

Long noncoding RNAs (lncRNAs) are pivotal regulators of genome structure and gene expression. LncRNAs can directly interact with chromatin-modifying enzymes and nucleosome-remodeling factors to control chromatin structure and accessibility of genetic information. Moreover, lncRNA expression can be controlled by chromatin-remodeling factors, suggesting a feedback circuit of regulation. Here, we discuss the recent advances of lncRNA studies, focusing on the function and mechanism of lncRNA–chromatin interactions.

Long noncoding RNAs (lncRNAs), which are defined as non-protein coding transcripts longer than 200 nucleotides, have emerged as important regulators of cell physiology and pathology. Transcriptome profiling has identified an increasing number of lncRNAs with tissue-specific expression; however, only dozens of lncRNAs have been characterized in vivo with regard to the mechanisms of action.^1-7 The majority of lncRNAs remains unknown for their biological functions, roles in diseases, and precise mechanisms. One unique feature of lncRNAs is their biochemical abilities to interact with a wide range of molecules, and through specific RNA functional domains lncRNAs can form a variety of RNA–RNA, RNA–DNA, or RNA–protein complexes,⁸ conferring functional diversities to lncRNAs. Recent discoveries of lncRNAs reveal a broad association of lncRNAs with the epigenetic machinery to control chromatin structure and gene expression. LncRNAs can directly interact with many histone- and DNA-modifying enzymes to participate in covalent modifications of histones or DNA. Furthermore, an lncRNA was recently found to be capable of modulating the non-covalent, ATP-dependent chromatin remodeling process,⁹ indicating an extensive role of lncRNAs in chromatin regulation. The mechanisms of how lncRNAs control chromatin by covalent modifications are previously reviewed.^10-12 Here, we discuss the recent progress of lncRNA studies, with an emphasis on the mechanism and function of lncRNA that modulates chromatin remodeling. We also discuss how lncRNA expression is regulated and how lncRNA–chromatin interactions can be used to design new therapeutic strategies for human diseases.

Ongoing characterization of long non-coding RNAs

The genome that encodes lncRNAs contains sequences that feature promoter, initiation codons, termination sites, and splice sites.^13,14 A recent computational biology study concluded that human cells contain at least 91,013 expressed transcripts, 68% of which (58,648) were classified as lncRNAs.¹⁵ The putative lncRNA promoters are enriched with histone 3 lysine 4 trimethylation (H3K4me3), RNA polymerase II (Pol II), and DNase I hypersensitivity sites,¹⁵ suggesting that lncRNA expression is actively regulated. This is consistent with the tissue- or developmental stage-specific expression of many lncRNAs. Like the mRNAs (mRNAs), lncRNAs are produced by independent transcription unit predominantly through the polymerase II complex, and most lncRNAs are polyadenylated.¹⁶ The lncRNA genomic loci harbor histone modifications, and the RNA transcripts exhibit alternative splicing that uses splicing signals (GT/AG) with exon and intron lengths comparable to those of mRNAs.¹⁶ However, in terms of exons number, a large fraction (42%) of lncRNAs consists of 2 exons in contrast to that (6%) of protein-coding genes.¹⁶ Although lncRNA transcripts and promoters are more conserved evolutionarily relative to random control regions,¹⁵ the degree of conservation of lncRNAs is less than that of mRNAs, and lncRNAs usually do not contain open reading frames that have cross-species mutations.^17,18 This generally low conservation of lncRNA primary sequence, however, doesn't preclude potential conservation of secondary or tertiary structure or the presence of conserved functional domains.¹⁹ Indeed, approximately 1% of lncRNA genomic loci harbor ultra-conserved elements, defined as DNA longer than 200 nucleotides that have nearly identical sequences across multiple species,¹⁵ suggesting the presence of lncRNA functional domains. Despite the modest overall conservation, 7% of the lncRNAs are found to overlap with disease-associated SNPs, indicating an association of these lncRNAs with diseases.¹⁵

LncRNAs are found in all subcellular fractions, with a subset of lncRNAs enriched in the nuclei, while the majority of lncRNAs located in the cytosol or in the ribosomal fractions.²⁰ Given that lncRNAs can functionally interact with many proteins, the association of lncRNAs with ribosomes could not be simply perceived as representing their protein-coding tendency. However, many lncRNAs do contain small open reading frames, and an emerging trend in the field is to explore the peptide- or micropeptide-coding potential of the RNA transcripts that were annotated as lncRNAs.^21,22 This aspect of lncRNAs is largely obscure, given the difficulty of identifying open reading frames that encode small peptides. Recent genome-wide and peptideomic studies revealed hundreds of small peptides that are encoded by annotated lncRNAs in vertebrates.^21,22 Functions of some small peptides encoded by “lncRNAs” are known. For example, a skeletal muscle-specific lncRNA encodes a 46-amino acid micropeptide Myoregulin (MLN), which is critical for inhibiting the membrane pump SERCA to regulate calcium uptake into the sarcoplasmic reticulum for muscle relaxation.²³ Another example is ELABELA, a 32-amino acid peptide hormone encoded by a previously annotated lncRNA.²⁴ During early cardiovascular development ELABELA transmits developmental signals from ectoderm to endoderm, and then to mesoderm through the putative Apelin receptor APLNR and Gata5- and Sox7-mediated pathways.²⁴ Therefore, for annotated lncRNAs that contain one or more open reading frames, efforts should be made to distinguish their functions resulting from the RNA molecules or from the translated small peptides.

LncRNAs as partners of chromatin-modifying enzymes

As previously reviewed,^10-12 lncRNAs can regulate gene expression through their interactions with chromatin-modifying enzymes, which catalyze covalent changes of histones or DNA on the chromatin to affect the expression of genetic information. LncRNAs are known to associate with many histone- or DNA-modifying enzymes, including Polycomb Repressive Complex (PRC),²⁵ MLL/TrxG complex,²⁶ histone demethylase LSD1,²⁷ DNA methyltransferase DNMT1,²⁸ and DNA demethylation regulator GADD45a.²⁹ Although the detailed biochemical mechanisms remain to be determined, lncRNAs may act through 2 different mechanisms to covalently modulate chromatin. First, lncRNAs can directly bind to chromatin-modifying enzymes, thus serving as a guide to anchor chromatin modifiers to targeted genomic regions or functioning as a decoy to sequester chromatin modifiers from specific genomic sites. A recent study showed that the lncRNA TARID (TCF21 antisense RNA inducing demethylation) acts by coupling with GADD45A to direct the DNA demethylation machinery to specific gene loci in cancer cells to regulate gene expression.²⁹ Second, lncRNAs can be incorporated into the chromatin-modifying complex and function as part of the complex or as a scaffold to assemble the complex for chromatin modification. In the fruit fly, for example, the lncRNA roX2, after being structurally remodeled by an RNA helicase MLE1,^30,31 becomes incorporated into the male-specific lethal (MSL) protein complex, which then recruits a histone acetyltransferase (MOF) to induce histone acetylation and activate gene transcription in the male X chromosome.^32,33 The lncRNA Xist provides another interesting example. Xist is bound by ATRX protein to promote PRC2 loading onto Xist and subsequent spreading of PRC2 along the X chromosome.³⁴ PRC2 then causes widespread H3K27 trimethylation and gene inactivation in the X chromosome.

LncRNAs as partners of ATP-dependent chromatin-remodeling factors

Besides covalent modifications of DNA and histones, active remodeling of the nucleosome composition and position provides another route for regulating chromatin and gene expression. However, it remains largely unaddressed whether lncRNAs are involved in the nucleosome remodeling process. Our recent study shed some light on how lncRNA may regulate nucleosome remodeling and gene expression through the ATP-dependent chromatin-remodeling complex.⁹

We identified a cluster of alternatively spliced, cardiac-specific, and nuclei-enriched lncRNAs that are associated with myosin heavy chain gene Myh7 in mice. The myosin heavy chain associated RNA transcript was named Myheart (Mhrt).⁹ Mhrt can protect the heart from stress-induced cardiac hypertrophy and failure through its direct inhibition of Brg1 chromatin function. Brg1 is an ATPase catalytic subunit of the SWI/SNF-like BAF chromatin-remodeling complex, and it is essential for the development of pathological cardiac hypertrophy³⁵ (Figs. 1A and 2). Brg1 contains a SF2 RNA helicase domain, which can bind to chromatinized DNA and tether Brg1 to its genomic targets for nucleosome remodeling and gene regulation. The Brg1 helicase domain can also bind to Mhrt RNA with high affinity, thus enabling a competitive inhibition mechanism by which Mhrt sequesters Brg1 from the genomic DNA loci and inhibits Brg1's gene regulation. Mhrt provides the first example of lncRNA capable of inhibiting the function of an ATP-dependent chromatin-remodeling factor in vivo. This Mhrt–Brg1 interaction thus showcases a direct link between lncRNA and nucleosome remodeling. Besides restructuring nucleosomes, Brg1 is a central player in recruiting other epigenetic factors to modify chromatin at specific genomic loci. Brg1 recruits at least 4 other classes of chromatin-modifying enzymes—Parp, Hdac, G9a/Glp, and Dnmt3—to converge on the promoter of Myh6, a cardiac-specific molecular motor gene. Once recruited to the target Myh6 promoter, these chromatin-modifying enzymes act in concert to generate repressive chromatin that mediates stress-induced Myh6 gene silencing and contributes to the development of cardiomyopathy (³⁵ and Chang lab unpublished Data). Mhrt, by inhibiting Brg1 from targeting Myh6 promoter,⁹ is likely capable of inhibiting the Brg1-centered covalent modifications of histones and DNA, which include poly-ADP-ribosylation (catalyzed by Parp), histone deacetylation (by Hdac), H3K9 methylation (by G9a/Glp), and DNA methylation (by Dnmt3). Further testing of such additional epigenetic actions of Mhrt will provide new insights into how lncRNA can lead to a cascade of epigenetic changes through a key chromatin remodeler.

Figure 1.

Two models of how lncRNAs regulate chromatin remodeling. (A), The lncRNA Mhrt can sequester Brg1/BAF chromatin-remodeling complex from their genomic targets to inhibit nucleosome remodeling.⁹ The action is mediated by direct interactions between Mhrt and Brg1. (B), Pol V-transcribed lncRNA tethers SWI3B-containing SWI/SNF chromatin-remodeling complex to genomic loci to repress gene expression in Arabidopsis.³⁶ Such action is accomplished through a RNA-binding protein IDN2, which binds both lncRNA and SWI3B.

Figure 2.

Reciprocal regulation of Brg1 and Mhrt. Cardiac stress activates Brg1 to complex with HDAC and PARP to repress Myh6 and Mhrt transcription through 2 separate promoter elements (purple).^9,12,35 Mhrt, on the other hand, can inhibit Brg1's chromatin targeting.^9,12

Studies of Arabidopsis suggested a model in which lncRNAs recruit Swi/Snf ATP-dependent chromatin-remodeling complex to genomic sites to position nucleosomes and silence gene expression³⁶ (Fig. 1B). In Arabidopsis thaliana, lncRNAs that are transcribed by RNA polymerase V (Pol V) from genomic sites containing retrotransposons and repetitive DNA elements are usually associated with gene silencing.³⁷ Transcription of these lncRNAs facilitates local heterochromatin formation to repress gene expression.³⁷ The Pol V-transcribed lncRNAs can bind to IDN2 (an RNA-binding protein essential for transcription silencing),^38,39 which then binds to the SWI3B subunit of Swi/Snf complex to recruit the remodeling complex to the sites of lncRNA transcription to establish repressive chromatin environment.³⁶ Therefore, lncRNAs can indirectly recruit chromatin-remodeling complex to local genomic sites to reposition nucleosomes for gene repression.

LncRNA transcription can be controlled by ATP-dependent chromatin-remodeling factors

The expression of lncRNAs is under tight regulatory control, with many lncRNAs showing tissue- or developmental stage-specific expression.^15,40,41,42 Such specific expression of lncRNAs raises a critical question regarding how lncRNA transcription is regulated in a precise temporal and spatial manner. The regulation of Mhrt expression may shed some light on the control mechanism.⁹ In mice Mhrt is specifically expressed in the heart: it is present at low level in fetal cardiomyocytes, and its transcription increases as the heart matures from the fetal stage to the neonatal period and to adulthood. This expression pattern coincides with that of Myh6, a cardiac-specific molecular motor gene essential for heart function. Mhrt and Myh6 are down-regulated when the hearts of adult mice or humans are pathologically stressed, and such down-regulation of Mhrt and Myh6 is crucial for the development of hypertrophy and heart failure.^9,35 Mhrt and Myh6 share the same promoter, whose activity is cardiac-specific and bidirectional—transcribing Myh6 in one direction and Mhrt in the other (Fig. 2).^9,35 This bidirectional promoter harbors 2 distinct promoter elements to separately control the transcription of Myh6 and Mhrt,^9,35 and both promoter elements are repressed by the chromatin-remodeler Brg1^9,35 (Fig. 2). Brg1 has an expression pattern opposite to that of Mhrt and Myh6. Brg1 expression is activated in fetal hearts, silenced in adult hearts, but reactivated by cardiac stress in adult hearts.³⁵ Once reactivated in adult hearts, Brg1 forms a chromatin repressor complex with Hdac and Parp on the bidirectional promoter to simultaneously inhibit Myh6 and Mhrt, resulting in cardiomyopathy.^9,35 In the heart, Mhrt transcription is repressed by Brg1, whereas Mhrt RNA itself can inhibit Brg1's chromatin targeting.⁹ This reciprocal inhibition indicates the presence of a Brg1–Mhrt feedback circuit critical for controlling cardiac physiology and pathology.^9,35 Moreover, the Brg1–Mhrt interaction illustrates that the ATP-dependent chromatin-remodeling factor can control the expression of lncRNA under different pathophysiological conditions.

A recent yeast genetic screening identified a large number of putative lncRNA repressors, which include 8 genes encoding subunits of 4 different ATP-dependent chromatin-remodeling complexes Rsc (RSC1, RSC2, and HTL1), Isw2 (ITC1), Ino80 (IES2), and Swr1 (SWR1, ARP6, and YAF9).⁴³ These chromatin-remodeling factors repress over 250 antisense lncRNAs (chromatin remodeling-repressed antisense transcripts or CRRATs) to maintain normal expression of mRNA transcripts that overlap with those antisense lncRNAs.⁴³ Therefore, the highly conserved ATP-dependent chromatin-remodeling factors may serve as global lncRNA repressors.⁴³ The repression of antisense lncRNAs may be a common mechanism of mRNA regulation. This view is consistent with the observation that Brg1 represses the expression of Mhrt, which is antisense to Myh7, to promote the expression of Myh7 mRNA in pathologically stressed hearts.^9,35 Future investigations are needed to pinpoint how chromatin-remodeling complex coordinates with other epigenetic regulators and transcriptional machinery to modulate transcription of lncRNAs, which may form a foundation to develop novel therapeutics by manipulating the expression of disease-associated lncRNAs.

Future perspective: the translational value of lncRNA–chromatin biology

LncRNAs have emerged as an important class of molecules implicated in human diseases, including cancers, cardiovascular diseases, neural degenerative disease, as well as metabolic disorders.^2,4-7,12 The profound lncRNA–chromatin interface provides new opportunities to unlock the potential of this mechanism for diagnostic or therapeutic applications. First, lncRNAs are more tissue-specific in contrast to the chromatin-regulating or epigenetic machinery that tends to operate widely in many tissues. Targeting lncRNAs may therefore result in better tissue specificity with lower general toxicity. Current technology holds great promise for therapeutic manipulation of lncRNAs through siRNA interfering, antisense oligonucleotides, aptamers, ribozymes, or CRISPR-mediated RNA cleavage.^44-46 In addition, direct interactions between lncRNAs and chromatin-regulating factors can be used as an avenue for developing assays to screen for small molecules and large molecule biologics that augment or dampen the lncRNA–chromatin interactions. Finally, more and more lncRNAs have been discovered as disease-causal or as the hub of new pathways that contribute to disease development.^2,4-7,12 Such increasingly fast discoveries of disease-associated lncRNAs are offering us unprecedented opportunities to develop novel therapeutics through the lncRNA mechanisms. The lessons we have learned from targeting microRNAs and delivering RNA-related reagents over the last decade can be modified and transferred to the lncRNA field to realize the therapeutic potential of lncRNAs.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Funding

C.-P.C. is the Charles Fisch Scholar of Cardiology and was supported by National Institutes of Health (NIH; HL118087, HL121197, HL085345), the American Heart Association (AHA; National Scientist Development Award; Established Investigator Award 12EIA8960018), Lucile Packard Heart Center Research Program, Oak Foundation, Baxter Foundation, Children's Heart Foundation, March of Dimes Foundation (#6-FY11-260), Indiana University (IU) School of Medicine—IU Health Strategic Research Initiative, and the IU Physician-Scientist Initiative, endowed by Lilly Endowment.