Introduction

Translation mechanisms of eukaryotic genes containing open reading frames and producing functional proteins are well established. The remaining DNA, not encoding messenger RNA (mRNA) and, therefore, not leading to protein production, was formerly considered 'junk DNA'.1 However, technological advances have allowed improved characterization of what are now called 'noncoding RNAs', which contribute to the complexity of organisms—the number of coding genes does not differ between a worm and a human, whereas noncoding RNAs are far more numerous in the latter. Noncoding RNAs have emerged as powerful regulators of cellular and tissue function and, therefore, can be used as diagnostic and prognostic markers, potentially allowing stratified health care and therapeutic intervention.2,3,4,5 Long noncoding RNAs (lncRNAs) are among the several families of noncoding RNAs that have arisen. The number of lncRNAs in the human genome is estimated to be equal to, or even to exceed, the number of protein-coding genes.6 Initially considered 'dark matter of the genome',7 lncRNAs can now be regarded as critical epigenetic regulators of gene expression.8 Studies have revealed that lncRNAs have important roles in healthy and diseased hearts. In this Review, we discuss the current understanding of the roles of lncRNAs in cardiac development and ageing.

LncRNA classification and mechanisms

Definition

Analysis of the human genome first led to the identification of long sequences of noncoding RNAs, such as ribosomal (rRNA) and transfer RNA (tRNA). Subsequently, other types of long noncoding RNA sequences, not translated into proteins, have been discovered. These sequences resemble mRNAs in several ways—they are transcribed by RNA polymerase II, 5′-capped, spliced, and polyadenylated at the 3′ end—but they do not encode proteins and are generally expressed at low levels compared with those of protein-coding genes. Consequently, lncRNAs were first defined as mRNA-like noncoding RNAs.7

Interest in noncoding RNAs greatly increased after the discovery by the ENCODE consortium that <2% of the human genome is transcribed into protein-coding RNA.9 High-throughput RNA sequencing and computational analyses substantially improved characterization of noncoding RNAs, and the definition of lncRNAs became more specific, with an arbitrary criterion of length. LncRNAs are defined as being noncoding RNA sequences of >200 nucleotides that regulate target-gene expression; can be categorized as being intergenic, intronic, bidirectional, enhancer, sense, or antisense; and can have a mechanism of action classed as being signal, decoy, guide, or scaffold (as described below).

Noncoding RNAs have been divided into two main groups: small noncoding RNAs that are <200 nucleotides, and lncRNAs containing >200 nucleotides.10 However, this threshold was arbitrarily defined according to protocols of RNA isolation. In 2011, Amaral et al. proposed to refine this definition as follows: lncRNAs are noncoding RNAs that can have a function as either primary or spliced transcripts, are independent of processing into known classes of small RNAs (microRNAs [miRNAs], piwi-interacting RNAs [piRNAs], and others) and exclude classic housekeeping families of RNAs (such as tRNAs and rRNAs).11 In the 2014 update of the NONCODE database, the number of entries reached 210,831 lncRNAs, with 95,135 in the human genome.12 Interestingly, many lncRNAs seem to be cardiac-specific, or at least cardiac-enriched.13,14,15,16

Classification

LncRNAs have most commonly been classified according to their genomic location. Six categories have emerged (Figure 1). Intergenic lncRNAs (lincRNAs) are located between two protein-coding genes; the majority of lncRNAs belong to this category. Intronic lncRNAs are located within introns of protein-coding genes. Bidirectional promoter lncRNAs are transcribed within 1 kb of promoters in the opposite direction from the protein-coding transcript. Enhancer lncRNAs (elncRNAs) are generally <2 kb and transcribed from enhancer regions of the genome. Sense lncRNAs are transcribed from the sense strand of protein-coding genes, and can overlap introns and part or all of the exon. Antisense lncRNAs are transcribed from the antisense strand of protein-coding genes, and can overlap an exon of the protein-coding gene in the sense strand, an intron, or both.

Figure 1: Classification of lncRNAs according to their genomic location.
figure 1

a | Intergenic lncRNAs (lincRNAs) are transcripts located between two protein-coding genes. b | Intronic lncRNAs are located in an intron of a coding gene. c | Bidirectional lncRNAs are located within 1 kb of promoters in the opposite direction from the protein-coding transcript. d | Enhancer lncRNAs (elncRNAs) are located in enhancer regions. e | Sense lncRNAs are transcribed from the sense strand of protein-coding genes and overlap one or several introns and exons. f | Antisense lncRNAs are transcribed from the antisense strand of protein-coding genes and overlap one or several introns and exons of the sense sequence. Abbreviation: lncRNA, long noncoding RNA.

PowerPoint slide

Full size image

Mechanism of action

In contrast to mRNAs or miRNAs, lncRNAs are poorly conserved between species, which does not imply a lack of function, but instead suggests a role in changing (or increasing) complexity of species.17 Although our understanding of the function of lncRNAs is only emerging, studies have revealed that lncRNAs are important regulators of multiple biological processes, both in the heart (Table 1) and in other organs. Whereas miRNAs are known to downregulate gene expression mostly by inducing mRNA degradation,18 regulation of gene expression by lncRNAs is under the control of more diverse mechanisms. A striking difference between miRNAs and lncRNAs is that the latter can either repress or activate gene expression. LncRNAs are mostly localized in the nucleus, where they regulate gene expression at the epigenetic level. In addition, a minority (∼15%19) of lncRNAs are present in the cytoplasm, where they regulate translation. In the nucleus, together with various nuclear proteins, lncRNAs have a structural role in forming and maintaining subnuclear domains. For example, the lncRNA NEAT1 has an essential role in the establishment of paraspeckles,20 a type of nuclear body. Quinozod and Guttman proposed a model in which lncRNAs establish a nuclear domain with target genes and bind proteins to form RNA–DNA–protein complexes.21 In simple terms, recruited proteins modify the chromatin, triggering chromatin conformation changes, and thereby lead to regulation of transcription. This model can be applied to all lncRNAs described so far. Chromatin modification can be achieved through at least four distinct mechanisms, as described by Wang and Chang (Figure 2),22 and as detailed below.

Table 1 LncRNAs with characterized functions in the cardiovascular system
Full size table
Figure 2: Mechanisms of action of lncRNAs.
figure 2

a | Signal lncRNAs regulate gene expression in response to various stimuli. For example, KCNQ1OT1 recruits EHMT2 and PRC2 to a specific action site and induces transcriptional silencing through chromatin methylation. b | Decoy lncRNAs sequester transcription factors, leading to repression of gene expression. MALAT1 traps SR proteins and regulates alternative splicing. c | Guide lncRNAs bind to ribonucleoproteins and guide their localization to specific target sites. FENDRR forms a complex either with PRC2 to silence the expression of target genes, or with TrxG/Mll proteins to activate gene expression. d | Scaffold lncRNAs interact with multiple components and activate or repress transcription. CDKN2B-AS1 (also known as ANRIL) binds to PRC1 and PRC2 and modulates chromatin methylation. e | Enhancer lncRNAs modulate expression of target genes by an interaction between enhancer and promoter regions of genes through chromosomal looping. Abbreviations: EHMT2, histonelysine N-methyltransferase EHMT2; HxKy, histone (number x) lysine (number y); lncRNA, long noncoding RNA; PRC, polycomb repressive complex; SR, serine/arginine; TrxG/Mll, trithorax-group/mixed lineage leukaemia.

PowerPoint slide

Full size image

Signal lncRNAs

Signal lncRNAs (Figure 2a) respond to various stimuli and, therefore, regulate transcription in a timely and spatially precise manner. One elegant example is the potassium voltage-gated channel subfamily KQT member 1 opposite strand/antisense transcript 1 (KCNQ1OT1). This is a signal lncRNA detected in an imprinted cluster, meaning that the two alleles of this gene are differentially expressed. Indeed, KCNQ1OT1 is expressed by the paternal chromosome and silenced on the maternal chromosome (the KCNQ1OT1 promoter is methylated on the maternal chromosome, but unmethylated on the paternal chromosome).23 The KCNQ1OT1 promoter is located in intron 10 of the KCNQ1 gene and is transcribed in the opposite direction to KCNQ1.24 KCNQ1OT1 recruits histone methyltransferases, such as G9a and polycomb repressive complex (PRC) 2, to the site of action in chromatin.25 G9a histone methyltransferase triggers dimethylation of lysine H3K9,26 and PRC2 triggers trimethylation of lysine H3K27.27 In this way, KCNQ1OT1 induces transcriptional silencing of imprinted genes through chromatin remodelling. This silencing can occur in cis for genes close to KCNQ1OT1 (CDKN1C, KCNQ1, SLC22A18) or in trans for genes distant to KCNQ1OT1 (ASCL2, CD81).Cis regulation refers to lncRNAs that are located in close proximity to their target gene, whereas trans regulation refers to lncRNAs that are located distantly from their target gene. Such regulation occurs in the placenta during development, illustrating the notion of tissue-specific and temporal regulation of transcription.28 However, although KCNQ1 expression is regulated by KCNQ1OT1, its imprinting seems to be independent of KCNQ1OT1 during cardiac development.29

Decoy lncRNAs

Decoy lncRNAs (Figure 2b) sequester regulatory factors leading to repression of transcription. Regulatory factors are either RNA-binding proteins such as transcription factors and chromatin-modifying enzymes, or RNA sequences such as miRNAs. A typical example of decoy lncRNA is metastasis associated lung adenocarcinoma transcript 1 (MALAT1). Predominantly localized in the nucleus, MALAT1 controls alternative splicing by regulating the phosphorylation and distribution of serine/arginine splicing factors in nuclear speckle domains.30 Although this mechanism has not been described in the heart, MALAT1 has been shown to regulate endothelial cell function and vessel growth.31 Other lncRNAs known in the cardiac context might also have a decoy mechanism of action and regulate alternative splicing. These include myocardial infarction associated transcript (MIAT), a risk factor for myocardial infarction,32,33 and the antihypertrophic lncRNA myosin heavy chain-associated RNA transcript (MHRT).13 Interestingly, studies have shown that lncRNAs can also act as decoy for miRNAs. Cardiac hypertrophy-related factor (CHRF; AK048451)34 and cardiac apoptosis-related lncRNA (CARL; AK017121)35 trap miR-489 and miR-539, respectively, in cardiomyocytes.The lncRNA SENCR, which is associated with smooth muscle cell contractility, might also act as a 'sponge', which traps miRNAs via this mechanism.36

Guide lncRNAs

Guide lncRNAs (Figure 2c) form complexes with ribonucleoproteins and conduct their localization to specific target genes. For example, the lncRNA FOXF1 adjacent noncoding developmental regulatory RNA (FENDRR) forms a complex with PRC2, which binds to specific sites in FOXF1 and PITX2 promoters, and thereby inhibits the expression of these two target genes.37 Conversely, FENDRR can also interact with proteins that activate gene expression, such as trithorax-group/mixed lineage leukaemia proteins.37

Scaffold lncRNAs

Scaffold lncRNAs (Figure 2d) form complexes simultaneously with several molecular components, which act as transcriptional activators or repressors. For instance, the lncRNA CDKN2B-AS1 (also known as ANRIL) recruits and interacts with PRC1 and PRC2 leading to silencing of the INK4b-ARF-INK4a locus.38 Importantly, ANRIL has been described as a genetic risk factor for coronary artery disease,39,40 and its expression level is associated with left ventricular dysfunction after myocardial infarction.41 In addition, knock-down of ANRIL decreases the expression of ADIPOR1, TMEM258 (also known as C11ORF10), and VAMP3, which are important genes in the regulation of glucose and fatty-acid metabolism.42

Enhancer lncRNAs

A fifth mechanism of action has been described, involving maintenance by elncRNAs of the interaction between enhancer and promoter regions of target genes through chromosomal looping (Figure 2e).43 ElncRNAs have been identified in the cardiac context and are discussed below.

Importantly, most lncRNAs are multifunctional and can have several mechanisms of action leading to activation or repression of gene expression. For instance, KCNQ1OT1 can be considered both a signal lncRNA28 and a guide lncRNA for G9a and PRC2 methyltransferases.25 HOTAIR acts via several mechanisms, such as signal, decoy, and guide, as reviewed previously.16 The molecular mechanisms triggered by lncRNAs are not yet fully elucidated. Further research might lead to the discovery of new mechanisms and help to refine the definition and classification of lncRNAs, as well as their role in cardiac pathophysiology.

LncRNA functions in the heart

Investigators have profiled cardiac-expressed lncRNAs,15,44,45,46,47,48,49 and identified hundreds of differentially expressed lncRNAs during cardiac development and disease, but the characterization of the functional roles of these lncRNAs is a tremendous task. Extra-deep sequencing (>400 million reads) of murine hearts after myocardial infarction identified novel cardiac-specific lncRNAs, mostly associated with enhancers (elncRNAs) and specific developmental chromatin-transition states.44 Several associative approaches allowed the identified elncRNAs to be linked with biological functions, such as modulating linkage to chromatin states of several organs, regulating transition of chromatin state in embryonic stem cells differentiating into cardiomyocytes, and affecting physiological traits implicated in cardiac remodelling.16 Of note, a weak expression level (elncRNAs were identified by extra-deep sequencing) does not preclude a biological function. Instead, these specific elncRNAs are thought to be far upstream of the transcriptional mechanisms known to control cardiac development and remodelling, as exemplified by the elncRNA Novlnc6, which is upstream of both the important cardiac development transcription factor homeobox protein Nkx-2.5 and its regulator bone morphogenetic protein 10.

Other approaches were employed to discover the function of novel lncRNAs. For instance, chromatin states were used as a starting point for identifying the function of FENDRR37 and MHRT,13 which are both binding protein factors that dictate (cardiac) chromatin signatures. LncRNA–miRNA–mRNA co-expression networks can also be used to associate lncRNAs with potential regulatory functions in the heart.44,45,49 Despite these attempts to characterize the function of lncRNAs in the heart, the regulation of cardiac pathways by lncRNAs is still poorly understood (Figure 3). Several lines of evidence indicate that lncRNAs regulate cardiomyocyte metabolism,35,42 hypertrophy,13,34 differentiation, and proliferation.14,37 In endothelial and smooth muscle cells, lncRNAs are involved in the regulation of migration and differentiation.31,36,50 In fibroblasts, lncRNAs control telomere structure,51,52 and are associated with senescence.53 Whether lncRNAs participate in the regulation of inflammation and fibrosis—two hallmarks of cardiac remodelling—remains to be determined. Not all differentially expressed lncRNAs might be functionally important, but their unique association with chromatin states and enhancers suggests fundamental signalling roles.

Figure 3: Regulation of cardiac pathways by lncRNAs.
figure 3

LncRNAs are expressed in various cell types and target multiple genes involved in important functions of the heart. *These lncRNAs have been shown to regulate these processes in the heart in vivo. The list of lncRNAs in this Figure is not exhaustive. Abbreviation: lncRNA, long noncoding RNA.

PowerPoint slide

Full size image

LncRNAs in cardiac cell fate

Cellular processes, such as proliferation and differentiation, control specialization during development. Specific patterns of cardiac gene expression govern cell fate and behaviour.54 These integrated transcriptional programmes are mediated by cis-regulatory elements. In particular, enhancers control spatial and temporal gene expression during embryogenesis.55 The mammalian genome contains >1 million enhancers, which are selectively activated by transcription factors.56,57 In the heart, several lines of evidence indicate that cardiac transcription factors act as enhancers in a combinatorial manner to regulate expression of cardiogenic gene networks.58,59,60 Lineage-specific transcription factors bind to enhancer elements, recruit co-activators and chromatin modifiers, and elicit epigenomic reprogramming and reorganization of the genome 3D architecture.

The transcriptional machinery has been found to involve enhancers and produce elncRNAs. Importantly, the expression of elncRNAs correlates with the activation of the enhancers, and transcription at enhancers is crucial for enhancer activity.61,62 Fetal cardiac enhancers are transcribed during development and their expression correlates with an active state,15,63,64 which strongly supports a role for elncRNAs in cardiac cell fate commitment. For example, activity at mm85, an enhancer controlling expression of its neighbouring gene MYOCD, produces an associated transcript.63 In a more global approach, 834 heart-specific intergenic regions associated with RNA transcription were identified.64 The majority mapped to enhancers and were enriched close to genes implicated in heart development. In addition, using deep RNA sequencing and ab initio reconstruction, hundreds of developmental enhancer-associated lncRNAs have been identified from differentiating embryonic stem cells.63 These transcripts undergo state transition during cardiac differentiation. Interestingly, the vast majority of newly identified lncRNAs that are modulated in adult stressed hearts are also associated with a heart-specific enhancer.44 These heart-specific elncRNAs undergo stage-specific chromatin state transitions during cardiogenic differentiation, as demonstrated using publicly available chromatin immunoprecipitation (ChIP)-sequencing data generated from embryonic stem cells differentiating into cardiomyocytes.44,65

Together, these findings support a role for elncRNAs in cardiac specification and differentiation. For example, lncRNAs associated with active enhancer states in mesodermal precursors or in cardiac precursor cells are implicated in cardiac mesoderm specification. However, lncRNAs shown to be active in cardiomyocytes are likely to be involved in cardiac differentiation and maturation. At the cellular level, knockdown of cardiac elncRNAs reduces expression of the neighbouring cardiac genes, consistent with cis-regulatory mechanisms.44,63 In undifferentiated precursors, elncRNA-dependent modulation of target genes is thought to control a cardiogenic programme. By contrast, in differentiated cells such as cardiomyocytes, these transcripts help to maintain the cell specialization via regulation of important identity genes.

The implication of lncRNAs in embryonic stem cell specification has been investigated using a loss-of-function approach.8 In particular, some lncRNAs were found to control commitment into the mesoderm through repression of nonappropriate cell fates. Although not directly associated with an enhancer sequence, BVHT is an example of a lncRNA with functional roles at different stages of differentiation.14 BVHT is expressed in cardiac mesoderm and is highly present in embryonic stem cells and cardiomyocytes. Depletion of BVHT in embryonic stem cells does not affect pluripotency, but severely impairs their capacity to produce differentiated cardiomyocytes.

Another lncRNA involved in cardiac lineage commitment is FENDRR.37 This lncRNA controls differentiation of tissues derived from the lateral mesoderm, which gives rise to the ventral body wall and heart. Consistently, disruption of FENDRR expression in developing embryos is lethal, in part as a consequence of heart failure. In addition, increased expression of Nkx-2.5 and GATA-6, two important regulators of cardiac lineage commitment, is observed in FENDRR mutant embryos.37 FENDRR is exclusively expressed in cardiac mesoderm progenitors during development, that is, in eomesodermin-positive cells.37

Taken together, these findings strongly support a role for lncRNAs in the control of stage-specific differentiation of pluripotent stem cells and cardiac precursor cells into cardiomyocytes. This new class of molecules provides potential targets to modulate progenitor cell fate and possibly induce regeneration in damaged hearts.

LncRNAs in cardiac development

Because many lncRNAs function as epigenetic modulators of gene expression,8 genome-wide RNA sequencing (RNA-Seq) is increasingly employed as a method of cataloguing noncoding RNA expression and simultaneously uncovering potential functional effects on coding RNA targets. Comprehensive RNA-Seq66 has been used to define the lncRNA expression signature of developing hearts at embryonic days ∼E8.2547 and ∼E13.5.15 Comparison of E8.25 mouse embryo RNA-Seq results to the 2,240 annotated lncRNAs in the ENSEMBLE database identified 1,378 lncRNAs, 225 (16%) of which had tissue-specific expression at this early developmental stage.47 More than 100 annotated and newly described putative lncRNAs had a 'heart-specific' expression pattern, and multiple examples of divergently expressed lncRNA–mRNA pairs were identified, suggesting functional relationships.47

In a study with a more cardiac focus, the lncRNA and mRNA expression signatures of developing hearts were evaluated by comparing RNA-Seq results from E13.5 mouse hearts, healthy adult mouse hearts, and actively hypertrophying adult mouse hearts (1 and 4 weeks after transverse aortic banding).15 By aligning to a curated list of ∼3,000 lncRNAs and comparing to adult liver and keratinocyte (skin) RNA-Seq data, 321 cardiac-expressed lncRNAs were identified, 117 of which were enriched at least threefold in heart; several of these cardiac-enriched lncRNAs have been referred to as 'cardiac-specific'.14 Two findings from this study are notable. First, the comparison between developing embryonic heart and quiescent adult heart showed major differences in lncRNA expression signature, whereas the comparison between quiescent and actively hypertrophying adult hearts showed far fewer lncRNA differences, consistent with postulated central roles for lncRNAs in tissue and organ development.67 Second, the lncRNAs that were highly expressed in embryonic hearts tended to have functional roles in pathways regulating transcription, including nuclear factor NF-κB and CREB. The investigators have postulated that increased lncRNA expression and counter-regulation of target-coding genes might limit the capacity of differentiating cells to revert to the multipotent state, whereas activation of cardiac-specific genes by conventional transcription factors provides the impetus for cardiomyocyte specification.15,68

Deep RNA sequencing has identified hundreds of lncRNAs thought to be involved in cardiac development. However, caution is warranted before assigning or accepting causal roles for lncRNAs, especially novel lncRNAs classified on the basis of computational analyses without biological validation.

LncRNAs and chromatin regulation

LncRNAs have crucial roles in regulating cardiac chromatin structure during heart development and pathological remodelling.13,14,37 In mouse embryos, Bvht14 and Fendrr37 act with histone-modifying enzymes to structure chromatin of gene promoters whose activities are required for heart development. In adult mice, the cardiac-specific lncRNA Mhrt inhibits the function of a pathogenic chromatin-remodelling factor Brg1 to maintain cardiac function and protect the heart from stress-induced failure.13

Bvht is thought to function by sequestering polycomb protein Suz12 from the promoters of cardiac lineage-determining genes (Hand1, Hand2, Mesp1, Nkx2-5, and Tbx20).14 Suz12 is a core subunit of PRC2 that catalyses trimethylation of histone lysine H3K27, providing a repressive histone mark (H3K27me3) that silences gene expression.69 Reduced Suz12/PRC2 promoter occupancy results in derepression of those genes that are critical for lineage determination. In cultured embryonic stem cells, Bvht functions as a decoy for PRC2 to reduce repressive chromatin marks and, therefore, activate lineage-specifying genes for the heart. Whether Bvht has similar roles in vivo requires further investigation.

In Fendrr-null embryos, the proximal Foxf1a was ectopically expressed, suggesting that Fendrr acts in cis to regulate gene-enhancer function.37 Furthermore, Fendrr binds with both the repressive complex PRC2 and the activating complex trithorax-group/mixed lineage leukaemia protein to control chromatin and gene expression for mesodermal and cardiac specification.37 Unlike Bvht that decoys PRC2, Fendrr helps to anchor PRC2 on the promoters of Foxf1 and Pitx2, two transcription factors essential for early cell fate decision, suggesting that Fendrr and Bvht might counteract each other to control the extent of PRC2 occupancy on promoters of cardiac lineage-determining genes, thereby tailoring heart development.

Mhrt transcripts, localized in the nuclei of cardiomyocytes, are weakly expressed in fetal hearts, but are abundant in mature adult mouse hearts.13 Cardiac stress suppresses Mhrt expression, leading to pathological hypertrophy and heart failure. Transgenic restoration of Mhrt to its pre-stress level prevents the heart from developing pathological hypertrophy. Mhrt inhibits the chromatin function of Brg1, an ATPase subunit of the BAF chromatin-remodelling complex that is activated by stress to induce cardiac hypertrophy.70 Mhrt directly binds to the ATPase/helicase domain of Brg1 and prevents Brg1 from recognizing its genomic targets, including the promoters of α-myosin and β-myosin heavy chain (Myh6 and Myh7) that have critical roles during cardiac hypertrophy. The helicase domain of Brg1 binds to Mhrt and chromatinized DNA promoter (Myh), allowing competitive inhibition of the genomic targeting of Brg1 by Mhrt. Interestingly, the stress-induced Mhrt repression is mediated through Brg1, which is activated in the stressed heart to repress the Mhrt promoter. Such reciprocal inhibition of Mhrt and Brg1 constitutes a feedback Mhrt–Brg1 circuit critical for maintaining the homeostasis of cardiac epigenome and function (Figure 4). Perturbation of the Mhrt–Brg1 circuit by pathological stress results in gene reprogramming and cardiomyopathy. Brg1 and Mhrt are both highly conserved in humans.13,70 As in mice, the expression of BRG1 is activated, whereas that of MHRT is repressed in human diseased hearts.13,70 The reciprocal MHRT–BRG1 changes in human hearts suggest a conserved lncRNA–chromatin mechanism of cardiomyopathy. These studies provide a new paradigm of lncRNA–chromatin regulation in heart development, differentiation, and disease. However, the poor conservation of some lncRNAs cautions against direct extrapolation of animal experimentations to human heart development.

Figure 4: Mhrt inhibits chromatin targeting and gene regulation by Brg1.
figure 4

a | Working model of a Brg1–Mhrt feedback circuit in the heart. Brg1 represses Mhrt transcription, whereas Mhrt prevents Brg1 from recognizing its chromatin targets. Brg1 functions through two distinct promoter elements to repress Myh6 and Mhrt expression bidirectionally. b | Molecular model of how Brg1 binds to its genomic DNA targets. Brg1 helicase (D1D2) binds chromatinized DNA, the C-terminal extension (CTE) binds histone H3, and bromodomain binds acetylated histone H3 or H4. Permission obtained from NPG © Han, P. et al. A long noncoding RNA protects the heart from pathological hypertrophy. Nature 514, 102–106 (2014).

PowerPoint slide

Full size image

LncRNAs in cardiac ageing

The constant increase in life expectancy together with low birth rates has resulted in a dramatic shift in the median age of the population in developed countries over the past few decades, and this trend is predicted to continue. Ageing is a highly complex and poorly-understood phenomenon that involves a decline in function on both the cellular and organ levels, leading to increased incidence of prevalent diseases such as cardiovascular disease. The pathways underlying age-associated diseases are still poorly understood, and specific diagnostic and therapeutic approaches are lacking for this growing demographic of the global population.

Given that the role of noncoding RNAs in cardiovascular ageing has been summarized previously,71 we discuss important lncRNAs with broad importance in the general biology of ageing, such as those affecting telomeres. The telomeric lncRNA TERRA (telomeric repeat-containing RNA) has been implicated in modulating the structure and processing of deprotected telomeres.51 Indeed, TERRA is of critical importance in the regulation of the telomeric DNA damage response.52 In one RNA-Seq study, lncRNA expression was compared between proliferating, early-passage, 'young' human diploid WI-38 fibroblasts and those expressed in senescent, late-passage, 'old' fibroblasts (PDL 52).53 Among the novel senescence-associated lncRNAs, senescence-associated lncRNA 1 (SAL-RNA1) was found to delay senescence. Many of these senescence-associated lncRNAs are now awaiting functional characterization and some might emerge as new age-specific drug targets.53

A MALAT1 loss-of-function genetic model indicated that MALAT1 is not essential for mouse prenatal and postnatal development.72 Furthermore, depletion of MALAT1 does not affect global gene expression, splicing factor level and phosphorylation status, or alternative pre-mRNA splicing. However, among a small number of genes that were dysregulated in adult MALAT1 knockout mice, many neighboured MALAT1, indicating a potential cis-regulatory role of MALAT1 gene transcription.72 Interestingly, inhibition of MALAT1 in vivo by oligonucleotides reduced vascularization, pointing to MALAT1 as an interesting target to manipulate angiogenic processes.31

The lncRNA H19 has also been associated with ageing; H19 gives rise to a long noncoding mRNA that is a precursor of several miRNAs that negatively affect cell proliferation.73 In addition, H19 is involved in cellular senescence regulation.74 Therefore, H19 might be an interesting target for manipulating ageing-associated disease processes. For example, multipotent germline stem cells have been found to be hypomethylated at the H19 locus, and also to express high levels of telomerase.75 However, detailed reports of the effects of H19 manipulation on biological ageing are currently lacking. With an ageing population, the burden of cardiovascular disease can only rise. LncRNAs are implicated in cellular senescence, but whether lncRNAs might constitute a novel class of 'anti-ageing drugs' is an appealing possibility that remains to be addressed.

LncRNAs as therapeutic targets

When considering the therapeutic potential of a new group of biologicals such as lncRNAs, some crucial questions need to be addressed. What is the precise biological role of these lncRNAs in each cardiac or vascular disease? Is it possible to inhibit or mimic their function pharmacologically, and is it safe to do so? What could the adverse effects be, depending on their function in other organs? Which is the best animal or in vitro model to perform preclinical studies, and how can the data be translated to humans?

To have an incremental value in the treatment of cardiovascular diseases in addition to the existing inhibition of neurohumoral activation, a new lncRNA-based drug should target specific molecular processes, such as inflammation, angiogenesis, fibrosis, or cell growth. Promisingly, individual lncRNAs seem to have distinct biological functions in these processes (Figure 3, Table 1) and, therefore, inhibition of their function might lead to novel treatments.

Inhibition of lncRNAs can be achieved using 'gapmers', which are frequently used to block lncRNA function. Gapmers are potent antisense oligonucleotides that inhibit specific nuclear targets. RNA–DNA heteroduplexes formed after sequence-specific binding of antisense oligonucleotides to their target lncRNA are cleaved by the enzyme RNase H. The latter is a ubiquitous enzyme found in both the nucleus and the cytoplasm of all cells, and hydrolyses the RNA of RNA–DNA heteroduplexes. Consequently, gapmers are potent inhibitors of their nuclear target lncRNAs.76 Some questions need to be answered before any application of lncRNA inhibition can become a reality in human diseases. Which combination of gapmers is best to obtain sufficient tissue and cell penetration, is safe, and has the desired half-life? What about alternatives to gapmers, such as aptamers (single-stranded DNA or peptide molecules that can bind to molecular targets such as proteins with high selectivity and affinity)? What are their adverse effects? Alternatively, given that lncRNAs, like miRNAs, can also be protective against disease development, is mimicking of lncRNA function possible? Currently, mimicking the function of noncoding RNAs in vivo remains a challenging and unachieved goal in RNA therapy.

The potential of targeting lncRNAs in cardiac or vascular diseases was first identified in animal models of angiogenesis or cell growth. Silencing of Malat1 reduced capillary growth not only in a mouse model of hind-limb ischaemia (with detrimental effects),31 but also in a rat model of diabetic retinopathy (with beneficial effects).77 Inhibition of lincRNA-p21 increased neointimal hyperplasia in an Apoe-knockout mouse model of carotid artery injury.78 In mouse hearts, repression of Mhrt drove the progression of cardiac hypertrophy to failure, and restoration to normal levels prevented cardiomyopathy.13 Animal studies have not yet revealed the therapeutic potential of lncRNAs in cardiovascular inflammation or fibrosis, but data on these applications are anticipated.

The results described above were obtained in rodent models, with only correlative human data on cardiac expression13 or blood levels79 related to disease progression. Given that lncRNAs—in contrast to well-conserved miRNAs—are poorly conserved between species (or even if conserved might differ in their molecular targets and precise biological role), translating animal findings to human application will be extremely challenging. Using relevant human cell lines, 3D ex vivo cell cultures of various cardiac cells might help to overcome these limitations of animal studies. Identification of human-specific lncRNAs, either as human orthologues from rodent-discovered lncRNAs as described in myocardial infarction,44 or being expressed during differentiation of embryonic stem cells into cardiomyocytes,63 is crucial for further development of lncRNA-based treatment in humans. Although the association between lncRNAs and heart disease suggests that they might be a novel family of therapeutic targets, many challenges and questions remain to be addressed before this aim can be achieved.

LncRNAs as cardiac biomarkers

The need to implement personalized cardiovascular medicine is generally agreed;80 however, available tools and methods still lack accuracy. Circulating biomarkers have greatly improved the diagnosis of acute cardiac conditions. Indicators of disease evolution would simplify clinical decision-making and allow health care to be tailored to individuals. Several families of biomarkers have emerged, with protein-based or peptide-based diagnostic assays now being used in patient care. RNA-based diagnostic assays have been developed, initially based on mRNA expression profiles,81 and subsequently circulating miRNAs have been identified as cardiac biomarkers with great potential for personalized medicine.3 Spatial, temporal, and disease-associated regulation of expression48,53,54,55,56 suggest that lncRNAs might also be useful biomarkers (Table 2).

Table 2 LncRNAs with potential biomarker applications
Full size table

The association between ANRIL and the risk of coronary artery disease39,40 suggests that genetic polymorphism of lncRNAs might serve as a cardiac biomarker. MIAT is another candidate biomarker—genetic variation in this lncRNA confers susceptibility to myocardial infarction.32 Using isoproterenol administration in mice as an experimental model of heart failure, the expression of 32 lncRNAs was found to be regulated in the failing heart as well as in whole blood and plasma samples.82 This study supports the notion that the blood constitutes a reservoir of lncRNAs that can be used as cardiac biomarkers, particularly given that lncRNAs are present in plasma-derived exosomes.83 A study of plasma levels of MT-LIPCAR, a mitochondrial lncRNA that is a predictor of the development of left ventricular remodelling after myocardial infarction and death in patients with systolic heart failure, was the first proof of the principle that plasma lncRNAs might be used as biomarkers of prognosis in cardiovascular diseases.79 Another lncRNA, MGAT3-AS1, is upregulated in plasma samples from patients with acute kidney injury and is a predictor of overall survival.84

Expression profiles of lncRNAs in whole blood cells have also been shown to have biomarker utility. Levels of HIF1A-AS2, KCNQ1OT1, and MALAT1 were upregulated, whereas those of ANRIL were downregulated, in patients with myocardial infarction compared with healthy volunteers.41 Furthermore, ANRIL, KCNQ1OT1, MALAT1, and MIAT were univariable predictors of left ventricular dysfunction after myocardial infarction.41 Importantly, ANRIL and KCNQ1OT1 remained significant predictors of left ventricular dysfunction in multivariable analyses and provided an additive predictive value.41

LncRNAs are now accepted to have functional roles in cardiac development and disease and, therefore, are potential therapeutic targets, but further studies are needed to demonstrate their biomarker utility. Whether circulating lncRNAs are stable and both easily and reproducibly measurable in biological samples remains to be established. The incremental value of lncRNAs in addition to available markers will have to be evaluated. Also, the localization of lncRNAs in plasma compartments, whether they are actively secreted, and whether they can act as paracrine mediators, needs to be determined.

Conclusions

LncRNAs have emerged as important regulators of cardiac development and ageing. At an early stage of life, lncRNAs control the differentiation of pluripotent stem cells and cardiac precursors into functional adult cardiac cells. At a later stage, they regulate cellular senescence and other pathways involved in cardiac disease. Although hundreds of lncRNAs are associated with cardiac development and disease, their functionality and diagnostic, prognostic, and therapeutic utility remain to be determined.