Non-coding RNAs in cardiovascular diseases: diagnostic and therapeutic perspectives

Abstract

Recent research has demonstrated that the non-coding genome plays a key role in genetic programming and gene regulation during development as well as in health and cardiovascular disease. About 99% of the human genome do not encode proteins, but are transcriptionally active representing a broad spectrum of non-coding RNAs (ncRNAs) with important regulatory and structural functions. Non-coding RNAs have been identified as critical novel regulators of cardiovascular risk factors and cell functions and are thus important candidates to improve diagnostics and prognosis assessment. Beyond this, ncRNAs are rapidly emgerging as fundamentally novel therapeutics. On a first level, ncRNAs provide novel therapeutic targets some of which are entering assessment in clinical trials. On a second level, new therapeutic tools were developed from endogenous ncRNAs serving as blueprints. Particularly advanced is the development of RNA interference (RNAi) drugs which use recently discovered pathways of endogenous short interfering RNAs and are becoming versatile tools for efficient silencing of protein expression. Pioneering clinical studies include RNAi drugs targeting liver synthesis of PCSK9 resulting in highly significant lowering of LDL cholesterol or targeting liver transthyretin (TTR) synthesis for treatment of cardiac TTR amyloidosis. Further novel drugs mimicking actions of endogenous ncRNAs may arise from exploitation of molecular interactions not accessible to conventional pharmacology. We provide an update on recent developments and perspectives for diagnostic and therapeutic use of ncRNAs in cardiovascular diseases, including atherosclerosis/coronary disease, post-myocardial infarction remodelling, and heart failure.

Introduction

Key role of non-coding RNAs in health and disease

About 99% of the human genome do not encode proteins, but are transcriptionally highly active and give rise to a broad spectrum of ncRNAs with regulatory and structural functions. The observation of a steeply increasing fraction of non-coding RNAs (ncRNAs), which is in contrast to the modest increase in the number of protein-coding genes during evolution from simple organisms to humans (Figure 1), suggests an overwhelming role of the ncRNAs¹ in humans. ncRNAs such as microRNAs (miRs), small interference RNAs (siRNAs) and long non-coding RNAs (lncRNAs) are novel regulators of cardiovascular risk factors and cell functions and thus candidates to improve diagnostics and prognosis assessment. Beyond this, however, ncRNAs have now fundamentally expanded our spectrum of therapeutic options. With regard to disease pathogenesis, it has become evident from the Encyclopedia of DNA Elements (ENCODE) project² and other studies^3–10 that limiting analysis to protein-coding regions of the human genome is inadequate, since many non-coding variants are associated with important human diseases. Inclusion of the non-coding genomic elements in pathogenetic studies appears mandatory and comprehensive transcriptome mapping includes small and large ncRNAs in addition to the protein-coding genes. In the heart, miRs regulate post-transcriptional gene expression and have been shown to control cardiovascular development, inflammation, hypertrophy, fibrosis, and regeneration (see recent review^11–13). Of note are also miRs that regulate cardiac contractility (miRs 25 and 22), regeneration (miR-302-367 and miR99/100 family), inflammation (miRs 155 and 221/222), and fibrosis (miRs 21, 208b, and 125b), as well as vascular functions.^14–28

Figure 1.

Evolution of the non-coding human genome. (A) Whereas the number of protein-coding genes remains surprisingly similar from simple to complex species, it is the non-coding part of the genome that increases dramatically with morphologic complexity, reaching ≈99% of the entire genome in humans. Multiple evidence suggests important roles of ncRNAs transcribed from the non-coding genome in human health and disease. (B): Whereas a vast number of ncRNAs with grossly different structures (e.g. lncRNAs, miRs, circRNAs) have been identified, however, in the majority of these molecules their functions are essentially unknown (details in Supplementary material online, Figure S1).

Non-coding RNAs as diagnostic and therapeutic tools

Beyond their application as diagnostic and prognostic biomarkers (Section ‘Non-coding RNAs as biomarkers’), ncRNA can also be the targets (e.g. miRs, lncRNAs) or tools (e.g. siRNAs)^29–32 of novel therapeutic strategies (Figure 2). Thus, RNA interference (RNAi)-mediating siRNAs (‘RNAi triggers’) are highly versatile ‘general-purpose’ tools e.g. for the silencing of protein-encoding genes via targeting of their mRNA. In 2006, the Nobel prize of physiology or medicine was awarded to Dr Fire and Dr Mello ‘for their discovery of RNA interference—gene silencing by double-stranded RNA’^32,35 which is now rapidly explored as a novel therapeutic principle for cardiovascular disease. Several studies demonstrated therapeutic potential of organ-targeted RNAi based on viral vectors^36,37 or synthetic RNAs,^38,39 and therapeutic strategies based on the modulation of miRs.^40–45 An increasing spectrum of endogenous ncRNAs is employed for the development of novel therapeutic ncRNA tools optimized for specific therapeutic requirements.^46–51

Figure 2.

Endogenous non-coding RNAs as blueprints for RNA therapeutics. Non-coding RNAs may be addressed as therapeutic targets, but an increasing spectrum of endogenous ncRNAs are also employed as blueprints for the development of novel therapeutic tools. The spectrum of possible therapeutic targets has thus expanded beyond proteins, but also the therapeutic ‘toolbox’. One current topic is therapeutic RNA interference triggers (siRNAs) originally developed from endogenous siRNAs as blueprints, and made clinically applicable based on sophisticated chemical modifications and coupling to carriers/ligands for tissue targeting^33,34 (details in Supplementary material online, Figure S2).

These investigations are only the beginning of therapeutic exploration, since ≈10 000 small ncRNAs arise from the human genome forming diverse RNA structures ranging from miRs to circular RNAs (circRNAs).^52,53 These small ncRNAs are only part of a broader spectrum of ncRNAs including ≈16 000 lncRNAs up to many thousand basepairs in size.^54–59 For lncRNAs, classification is still in its infancy^60,61 and biological functions mostly unknown.^62–66 However, recent studies strongly suggest that they constitute fundamentally new therapeutic targets^67–70 with unusual RNA structures^71,72 to be addressed using new drug types. Their importance for human health and disease is highlighted by observation that both small and large ncRNAs rapidly evolved during primate evolution and are often primate-specific and cannot be studied in common animal models of human disease.^66,73 Overall, only a small part of the non-coding human genome has been investigated and explored with regard to possible therapeutic implications for cardiovascular medicine.

Search for non-coding RNAs of clinical interest

Because of the complexity of the non-coding genome, one may adopt a pragmatic heuristic approach directed to the identification of non-coding therapeutic targets in cardiovascular diseases (Figure 3). Insights from genome-wide phylogenetic studies, comprehensive sequencing work, and animal models of human cardiovascular diseases constitute an indispensable first step towards target identification. Importantly, the functions of the same ncRNA may differ between species, and most animal models differ significantly from the respective human disorder. Some ncRNAs do not even exist in humans⁷³ or have only distant orthologues, so that their possible pathogenic relevance in humans cannot be investigated in animals. Therapeutic proof-of-concept studies of ncRNAs deregulated in animal models are therefore not necessarily predictive for effects in humans.

Figure 3.

Search strategies for non-coding RNAs of clinical interest. Genome-wide genomic and transcriptomic mapping have revealed a huge number of novel non-coding transcripts and provided structural information regarding their gene loci, but for the vast majority there is as yet no functional assignment. Functional characterization is complicated by their often extensive processing. Prospecting of this plethora of ncRNA for possible clinical significance may be critically facilitated by state-of-the-art transcriptional mapping of cardiovascular patients, in order to obtain first functional hints to guide subsequent in-depth experimental investigations including therapeutic proof-of-principle studies.

It is therefore necessary to conduct screening studies for ncRNA of clinical interest also directly in humans, in particular in well-defined patient cohorts suffering from a carefully phenotyped cardiovascular disease of interest. Different principles build the basis of a successful screening project. First, relevant ncRNAs may be identified by genome-wide association studies (GWAS) indicating genomic loci conferring diseases risk via inherited mutations.^74–83 A second approach is comprehensive transcriptome mapping in diseased organs, which may reveal genomically encoded defects, but also transcriptome shifts developing during pathogenesis but not encoded at the DNA level.⁸⁴ Both approaches have already identified ncRNAs involved in cardiovascular disease pathogenesis. State-of-the-art analytical technologies enable both GWAS studies and comprehensive transcriptional mapping at high efficacy. However, some logistical and developmental aspects need to be considered that relate to fundamental limitations of these search strategies. Genome-wide association studies commonly addresses early processes in disease pathogenesis by analysing the genome at the DNA level of inherited mutations acting from childhood. Late processes during disease development, in contrast, are rather revealed by transcriptome mapping of already injured organs.⁸⁵ In addition, the selection of properly phenotyped patients, sufficient cohort sizes, and proper control groups often constitutes a leading problem in such screening projects. In a third approach, high-throughput functional screening for relevant ncRNAs in cell types of pathogenic relevance (cardiomyocytes,⁸⁶ vascular cells,^87,88 cardiac fibroblasts,^88–95 hepatocytes^96,97) has been successfully used to identify new ncRNA with diagnostic and therapeutic potential. Disease-associated ncRNAs identified by any of these approaches may yield biomarkers to assess the risk of disease progression, thus facilitating treatment decisions, or to predict response-to-therapy to optimally allocate limited resources. In addition, these ncRNAs may serve to develop and evaluate novel therapeutic strategies beyond optimal current therapies.

In the following, we discuss diagnostic and therapeutic perspectives of ncRNAs in cardiovascular diseases. Three clinical goals are running as guiding threads throughout the article. First, the identification of novel cardiovascular pathomechanisms and high-priority target diseases (Section ‘Non-coding RNAs as potential therapeutic targets in cardiovascular disease’). Second, the development of clinically useful RNA-based therapeutics and identification of patients most likely to benefit (Section ‘Clinical translation of RNA therapeutics’). Third, the employment of RNA diagnostics to improve prognosis assessment and individualized clinical patient management (Section ‘Non-coding RNAs as biomarkers’). These issues are highlighted again in a Summarizing Illustration: Delineation of current high priority diseases, patients most likely to benefit, status of technical feasibility of ncRNA therapies, and the current value of RNA diagnostics for prognosis assessment and patient management.

Non-coding RNAs as potential therapeutic targets in cardiovascular disease

Here, we review the identification of non-coding RNAs defining novel pathomechanisms, and important cardiovascular diseases significantly influenced by these novel processes.

MicroRNAs as therapeutic targets

Post-myocardial infarction adverse cardiac remodelling and dysfunction

MicroRNAs control important processes that contribute post-infarction injury and subsequent remodelling responses. For example, miRs can promote or inhibit cardiomyocyte cell death, regulate post-ischaemic neovascularization and control cardiac fibrosis (for an overview of miRs that control post-infarction repair see⁹⁸).

Cardiomyocyte cell death is induced by members of the miR-15 family, which are upregulated by myocardial ischaemia.⁹⁹ Inhibition of miR-15 family members by short, locked nucleic acid (LNA)-based anti-miRs, which target the seed sequence of most miR-15 family members, reduced infarct size after ischaemia–reperfusion injury by derepressing the anti-apoptotic protein Bcl-2, the mitochondrial protecting factor ADP-ribosylation factor-like protein 2, and the deacetylase SIRT1 in mice.^98,99 The feasibility of using miR-15 inhibitors has been documented also in pig models.⁹⁹ Myocardial infarction additionally induces the expression of the pro-apoptotic miR-34 family members. Inhibition of miR-34 family members by different types of anti-miRs reduced infarct size, and augmented the recovery of heart function after acute myocardial infarction (AMI) by increasing the expression of cardioprotective SIRT1 and PNUTS.^23,100 Inhibition of miR-15 or miR-34 family members also improved neovascularization after ischaemia.^23,101 Members of the miR-17∼92 cluster, specifically miR-92a, control neovascularization and cardiomyo-cyte cell death after ischaemia.^102,103 Pharmacological inhibition or genetic deletion of miR-92a increased capillary density and improved heart function after AMI in mice.^102,103 Furthermore, catheter-based delivery of anti-miR-92a or encapsulated antagomiRs reduced infarct size in a large animal cardiac ischaemia/reperfusion pig model.¹⁰⁴ Targets of miR-92a include integrin α5, KLF2 and SIRT1.^102,104,105miR-26a is increased after AMI and inhibits angiogenesis.¹⁰⁶ Administration of a LNA-based anti-miR against miR-26a induced robust angiogenesis, reduced myocardial infarct size, and improved heart function.¹⁰⁶ Mitochondrial function is controlled by miR-140, which reduces mitochondrial fission,¹⁰⁷ thereby impairing cardiomyocyte survival. Pharmacological inhibition reduced infarct size after AMI in mice.¹⁰⁷

On the contrary, there are cardioprotective miRs^108–111 including miR-210 which protects cardiomyo-cytes and vasculature. Intramyocardial injection of minicircle vectors carrying the miR-210 precursor reduced cell death and improved cardiac function and angiogenesis after AMI.¹⁰⁸

Cardiac hypertrophy and heart failure

One of the first reports about a miR involved in cardiac hypertrophy stem from the Olson laboratory, here miR-208 has been shown to be involved in hypertrophic signaling.¹¹² Inhibition of miR-133 with an anti-miR-133 oligonucleotide generated cardiac hypertrophy in vivo showing a role of miR-133 in cardiac hypertrophy as well.¹¹³ Other miRs that are regulated during pathological stress include the miR-212/132 family which becomes activated during heart failure (HF) in humans¹¹⁴ and animal models.⁴⁰ This miR family both regulates cardiac autophagy and hypertrophy in cardiomyocytes at least in part by modulation of transcription factor forkhead box O3. Inhibition of miR-21 prevented HF development in mice.⁴⁰ MicroRNAs also regulate intracellular calcium homeostasis, a process that is strongly deregulated during HF, thereby directly affecting cardiac contractility.^86,92

For few specific cardiomyopathies, the therapeutic potential of ncRNAs has been investigated. Thus, recent studies identified novel miR therapeutic targets in coxsackievirus-B3 (CVB3) myocarditis. MicroRNAs 155, 146b, and 21 are upregulated in acute CVB3-myocarditis.^115,116 Inhibition of miR-155^114,117,118, miRs 21 and 146b¹¹⁹ by systemically delivered anti-miRs attenuates cardiac inflammation and myocardial damage in CVB3 or autoimmune myocarditis in mice.

Atherosclerosis

Atherosclerosis underlying coronary artery disease (CAD), stroke, and peripheral vascular disease is a chronic inflammatory reaction of the arterial wall characterized by maladaptive responses of endothelial cells at sites with disturbed flow conditions and of immune cells, in particular macrophages accumulating lipids without resolving inflammation in a context of dyslipidaemia. A plethora of miRs and relevant targets has emerged in mouse models of atherosclerosis to intricately orchestrate crucial mechanisms, thus forming a basis for potential therapeutic strategies using anti-miRs or miR mimics.^120–122 An atheroprotective role in endothelial regeneration has been identified for the miR-126 strand pair. Depletion of miR-126-5p at predilection sites with altered flow under hyperlipidaemic stress limits the endothelial proliferative reserve and promotes lesion formation through derepression of DLK-1¹²³ which can be restored by miR-126-5p mimics. Furthermore, delivery of miR-126-3p by microparticles from apoptotic to recipient endothelial cells amplifies CXCL12 expression by repressing RGS16 to recruit proangiogenic cells supporting endothelial recovery,¹²⁴ a mechanism possibly impaired in CAD or diabetes.¹²⁰ Similarly, transfer of KLF2-induced miR-143/miR-145 by endothelial microvesicles to smooth muscle cell (SMCs) attenuates atherosclerosis by generating a contractile phenotype, as with miR-145 overexpression.^26,125miR-181b targeting importin-α3 is upregulated by laminar flow and limits endothelial inflammation and atherosclerosis by inhibiting NF-κB activation.¹²⁶ Conversely, miR-92a is upregulated by disturbed flow and modified lipoproteins to promote NF-κB-mediated endothelial inflammation and atherosclerosis by targeting KLF2/4, which is reversed by anti-miRs.¹²⁷ Likewise, the pre-ribosomal RNA-derived miR-712 is upregulated by altered flow to elicit endothelial inflammation and atherosclerosis by targeting TIMP3.¹²⁸

Macrophage subtypes show differential profiles of miR expression. Notably, miR-155 increases during atherogenesis or upon M1-type macrophage polarization, and promotes inflammatory activation and advanced atherosclerosis by targeting BCL6, a transcriptional regulator counteracting NF-κB, and by reducing cholesterol efflux.^129,130 Released from competing targets in M1-macrophages, miR-342-5p suppresses Akt1 to cause inflammatory activation and atherogenesis by upregulating miR-155 and forming a functional tandem for anti-miR strategies.¹³¹

The cholesterol efflux from macrophages to HDL is mediated by the ATP-binding cassette transporters ABCA1 and ABCG1. The ABCA1 3′UTR encompasses a wealth of binding sites for miRs including miR-33a/b, miR-19b, miR-144 and miR-148a.¹³² Beyond miR-33, several miRs targeting ABCA1/G1 can inhibit cholesterol efflux and have been implicated in atherogenesis.^120,121,132 In contrast, short-term antagonism of miR-33 in rodents and non-human primates significantly elevates plasma HDL, but overall the efficacy of miR-33 silencing in atherogenesis remains controversial.^{120–122,132}

Table 1 and Figure 4 provide short overviews of current miR-targeting cardiovascular therapies.

Table 1.

Translational studies addressing the therapeutic potential of non-coding RNAs

Investigation	Strategy	Technology	Reference
Clinical Trials
LDL-C reduction	Therapeutic RNAi targeting PCSK9 reduced LDL-C in non-human primates	Synthetic siRNA via nanoparticle carrier	133
LDL-C reduction	RNAi drug targeting PCSK9 reduced LDL-C in volunteers (intravenous siRNA administration in lipid nanoparticles)	Synthetic siRNA via nanoparticle carrier	134
LDL-C reduction	RNAi drug targeting PCSK9 reduced LDL-C in volunteers (subcutaneous siRNA administration conjugated to GalNAC)	synthetic siRNA conjugated to GalNAC	135
LDL-C reduction	ASO targeting PCSK9 reduced LDL-C in non-human primates	LNA-ASO via nanoparticle carrier	136
Lp(a) reduction	ASO targeting apolipoprotein(a) reduced Lp(a) in a randomised, double-blind, placebo-controlled clinical study	2ʹ-MOE-ASO via nanoparticle carrier	137–139
Transthyretin suppression	Clinical RNAi therapy for transthyretin amyloidosis inhibited hepatic synthesis of mutant TTR protein	Synthetic siRNA via nanoparticle carrier	140
Antiviral	RNAi-based antiviral treatment in non-human primates	Synthetic siRNA via nanoparticle carrier	141
Experimental Research
Vascular-targeted anti-miR	miR-92a inhibition of ischaemia/reperfusion injury, and improvement of re-endothelialization following vascular injury	LNA-ASO intracoronary injection (swine)	104,142
Cardiac-targeted RNAi	Cardiotropic silencing of Ca2+ cycle regulator phospholamban for the treatment of severe heart failure	Cardiac-targeted AAV9-vector-derived shRNA (rat)	36
Monocyte-targeted RNAi	Nanoparticle-encapsulated synthetic siRNA for silencing of monocytic CCR2	Non-viral, lipid nanoparticle-mediated siRNA delivery targeting particularly monocytes (mouse)	38
RNAi imaging in vivo	PHD2-shRNA followed by a hypoxia response element-containing promoter	Imaging of RNAi in space and time (mouse)	143
Cardiac RNAi	RNAi induced alloimmune tolerance in heart transplantation model TLR adaptor silencing	siRNA (rat)	144
Vasculature-targeted RNAi	RNAi to silence chymase increased plaque stability	Lentivirus-based RNAi	145
RNAi visualization	Visualizing lipid-formulated siRNA release and target knockdown	Development of strategies to improve the delivery of candidate RNAi drugs	146
Improvement of stem cells by RNAi	RNAi to enhance survival and function of transplanted cells	Enhancement of the efficacy of cardiovascular stem cell therapies	147
Allele-specific RNAi	Allele-specific RNAi for human induced pluripotency stem cell cardiomyocytes	Rescue of autosomal dominant-negative disorders	148
Safety issues
	Scientific and Regulatory Policy Committee: Biotherapeutics		149
	Scientific and Regulatory Policy Committee: Antisense Oligonucleotides		150
Emerging technologies
Gene silencing	siRNA vs miRNA for gene silencing	Comparison of therapeutic siRNAs and miRNAs	144,148,151,152
RNA engineering	High-throughput cellular RNA device engineering		153
RNAi visualization	Visualizing lipid-formulated siRNA release and target knockdown	To facilitate development of rational strategies to improve the cytosolic delivery of candidate drugs	146
RNAi-based functional profiling	RNAi-based functional profiling of loci from genome-wide association studies	RNAi to analyse 133 candidate genes in 56 loci identified by GWAS	154

ASO, antisense oligonucleotide; LNA, locked nucleic acid; RNAi, RNA interference.

Figure 4.

Cardiovascular RNA drugs address multiple organ systems. RNA drugs do not need to address heart (A) or vasculature (B) directly. Primarily liver-targeted RNA drugs (C) are the currently most successful development in the cardiovascular field. Another group of strategies addresses the immune system (D), in particular monocytes-macrophages. Carrier denotes a synthetic nanoparticle and/or receptor ligand employed to deliver an RNA drug to its tissue target.^{29,31,33,34,155–160} ‘Carrier’ is bound to and serves to stabilize the RNA drug within the circulation, and to endow it with at least partial selectivity for the target cells, in order to minimize side effects. ‘AAV9’ denotes a cardiac-targeting recombinant adeno-associated viral vector containing a genome from which the therapeutic RNA sequence is continuously transcribed (for transcript types see Figure 5). apo(a), apolipoprotein (a); CCR2, chemokine C-C motif receptor 2; CHAST, Cardiac hypertrophy associated transcript; PCSK9, proprotein convertase subtilisin/kexin type 9; PLB, phospholamban.

Long non-coding RNAs as potential therapeutic targets

Long non-coding RNAs research is challenging due to the fact that most lncRNAs, in contrast to miRs, are not conserved among species. Even when selecting the few being conserved, epigenetic effects and the processes of DNA and RNA assembly they influence may differ among species, making translation from rodents to humans challenging. Encouraging is the fact that lncRNAs have distinct functions in cardiovascular diseases (reviewed^170–176) and therefore may open new therapeutic opportunities.¹⁷⁷

Supplementary material online, Figure S1 depicts unsolved issues in translational lncRNA research. First, development of a more advanced classification^{54,56,60,178–183} of the vast number of lncRNAs detected in humans—based on structure, subcellular localization, intracellular processing, functional modules—is needed to facilitate rational development of lncRNA-targeting treatments. Second, a peculiar aspect of lncRNAs to be addressed is their often extensive and specific post-translational processing with differential subcellular localization of the products, yielding complex RNA processing systems with different functions of the products^69,184 which may be differentially addressable by drugs.¹⁸⁵

Cardiac hypertrophy and heart failure

Next to miRs, lncRNAs also play an important role in cardiac failure. The lncRNA Cardiac hypertrophy associated transcript (CHAST) is increased during cardiac hypertrophy in mice and humans and its inhibition both prevented and attenuated cardiac remodelling and HF.¹⁸⁶ In the heart, suppression of Myosin heavy-chain-associated RNA transcript (MHRT) accelerated progression of hypertrophy to failure,¹⁸⁷ whereas overexpression of CHAST induced cardiomyocyte hypertrophy. Gapmer-mediated CHAST silencing both prevented and attenuated pressure overload-induced pathological cardiac remodelling without evident early toxicological side effects.¹⁸⁶ Knockdown of lncRNA Cardiac mesoderm enhancer associated ncRNA (CARMEN) inhibits cardiac specification and differentiation in cardiac precursor cells. This occurs independently of miR-143 and -145 expressions despite the fact that those two miRs located proximal to the enhancer sequences. CARMEN interacts with SUZ12 and EZH2, two components of the polycomb repressive complex 2 (PRC2).¹⁸⁸

Atherosclerosis

Beyond miRs, lncRNAs may regulate atherosclerosis but their role and molecular mechanisms remain to be elucidated. As a strong genetic risk locus, the chromosome 9p21 (Chr9p21) region encodes the lncRNA antisense noncoding RNA in the INK4 locus (ANRIL). Its transcript expression correlates with the severity of atherosclerosis, and ANRIL regulates target genes in trans through Alu motifs in their promoters, leading to increased cell proliferation and adhesion but decreased apoptosis, relevant to atherosclerosis.¹²⁴ Conversely, lincRNA-p21 down-regulated in atherosclerotic plaques has been found to control neointima formation, repressing SMC proliferation and inducing apoptosis by enhancing p53 activity.¹²⁵ Intense investigations into the function on other miRs and lncRNAs yet to be identified are currently underway.

The potential of targeting lncRNAs was first described in models of angiogenesis or cell growth. Silencing of Metastasis associated lung adenocarcinoma transcript 1 (MALAT1) reduced capillary growth not only in a mouse model of hind limb ischaemia (detrimental),⁶⁸ but also in a rat model of diabetic retinopathy (beneficial).¹⁸⁹MALAT1-derived mascRNA is involved in innate immunity and viral myocarditis,¹⁹⁰ but appears to be dispensable in pressure overload-induced HF in mice.¹⁹¹ Next, inhibition of lincRNA-p21 aggravated neointimal hyperplasia in a carotid artery injury model in ApoE knockout mice.¹⁹²Smooth muscle and endothelial cell-enriched migration/differentiation associated RNA (SENCR) is a vascular cell-enriched, cytoplasmic lncRNA that seems to stabilize the smooth muscle cell contractile phenotype.⁸⁷ All of the above was obtained in rodent models, however, with only correlative human data on cardiac expression^186,187 or blood levels.¹⁹³

Since lncRNAs are poorly conserved between species, translation of animal findings to patients will be challenging. Whether the use of more relevant human cell lines, like inducible pluripotent stem cells may help to overcome these limitations of animal studies, remains to be determined. Identification of human-specific lncRNAs, either as human orthologues from rodent-discovered lnRNAs,¹⁹⁴ or being expressed during embryonic stem cell differentiation into cardiomyocytes,¹⁹⁵ will be important for further advance of lncRNA-based treatments in humans.

Clinical translation of RNA therapeutics

Several novel pathomechanisms influenced by ncRNAs, discussed above, cannot be therapeutically addressed using conventional pharmacology. ncRNA therapeutics offer new options to influence these hitherto inaccessible disease processes.

Versatility of therapeutic non-coding RNA structures

Small interference RNAs and related structures mediating RNAi and thus target gene silencing are the currently most advanced and frequently used types of ncRNA for therapeutic purposes, and part of a continuously expanding spectrum of therapeutic ncRNA tools developed from endogenous ncRNA as blueprints (Figure 2). Fundamentally different from DNA, RNAs are carrying information not only in their linear sequences of nucleotides (primary structure), but local nucleotide pairing creates secondary structures e.g. Hairpins, and interactions among distantly located sequences create tertiary structures.¹⁹⁶ In fact, this structural versatility needs to be considered for RNAs as therapeutic tools as well as targets. The plethora of RNA types, sequences, and structures created by evolution is a treasure trove of potential therapeutic tools and targets (Figure 2).

Classification of non-coding RNA therapeutics

Figure 5 provides an overview of currently available ncRNA therapeutics and their key properties: (i) sufficient stability or continuous in vivo production of therapeutic ncRNA, or the ability to redose if needed, (ii) high specificity of therapeutic ncRNA for the molecular target, (iii) proper targeting to the correct cells type or tissue by use of appropriate vectors plus transcriptional targeting, (iv) side effects induced by the therapeutic ncRNA itself or its delivery system need to be minimized, and (v) regulatability for certain applications.¹⁹⁷ Overall, RNA therapeutics may be chemically synthesized nucleic acids delivered using chemical delivery systems,^{33,34,198–200} or produced in vivo by recombinant viral vectors targeting different tissues.^201–209 Viral vectors have the capacity to act as long-term productive ‘RNA drug factories’, whereas chemical synthesis allows introduction of RNA modifications that cannot be generated biologically (details in Supplementary material online, Figure S3). Table 1 summarizes studies employing ncRNAs in clinical trials, preclinical animal models, or other experimental studies in vivo.

Figure 5.

Classification and key features of RNA therapeutics. For RNA drug production, two general approaches are used: in vitro chemical synthesis of backbone-modified RNAs, or in vivo drug transcription directly from a viral vector. For drug delivery in vivo, diverse synthetic systems have been developed, including nanoparticles and ligands with affinity for the target tissue, and also recombinant viral vectors as biological tools with appropriate tissue-targeting properties. Since high targeting efficacy and specificity is critical for therapeutic success, delivery system are often combined with ‘physical’ targeting employing catheter-based techniques reviewed recently.³⁷ Once the RNA drug has arrived in the target cells, two principles of action are currently in use: (i) RNA interference triggers as a highly versatile multipurpose tool for gene silencing.^{32–35,70,135,161,162} (ii) MicroRNA therapeutics for ablation^{33,34,163–168} or enhancement^{120–122,169} of miR functions. Apart from these differences, however, few common key characteristics will determine clinical usefulness, e.g. insufficient targeting or stability prevents efficacy, whereas lack of specificity or immune activation may cause severe side effects (details in Supplementary material online, Figure S3).

Challenges in translation

Clinical translation of these fascinating and far-reaching options faces challenges ranging from ncRNA drug design and delivery (Box 1) to regulatory issues (Box 2). Whereas animal studies are most valuable and indispensable for proof of principle, most of these studies use young otherwise healthy animals and thus rarely reflect the clinical reality. Taking HF as an example, few if any animal models mimic the human HF reality, a chronic systemic condition lasting years to decades. Models of chronic HF in animals are rare and only recently a model emerged which combines diabetes, obesity, hypertension, and leads to kidney dysfunction.^210,211 On the solid foundation of animal studies, few pioneering clinical trials were successful in demonstrating the technical and clinical feasibility of ncRNA therapeutic approaches.

Liver-targeted RNA drugs for the treatment of cardiovascular disease

Eminent recent examples of ncRNA therapies illustrate the important fact that cardiovascular RNA drugs do not need to address the heart or vasculature directly (Figure 4). On the contrary, primarily liver-targeted RNA drugs are the currently most successful development in the cardiovascular field. Thus, hepatic targeting of the proprotein convertase subtilisin/kexin type 9 (PCSK9) gene by intravenous injection of RNA drugs encapsulated into nanoparticles with high affinity for the liver achieved highly significant LDL cholesterol lowering.^134,212,213 An important recent development for optimization of liver delivery of PCSK9-siRNA has been the conjugation with GalNAC, allowing subcutaneous administration and a reduced dose and prolonged efficacy of the siRNA drug.^135,161 The ongoing clinical phase 2 trial ORION-1 evaluates prolonged dosing intervals as well as efficacy and safety of this approach in 501 patients. RNAi silencing of the transthyretin (TTR) gene in the liver as a novel treatment for TTR cardiac amyloidosis is also being further evaluated,¹⁴⁰ using improved liver-targeted strategies allowing for dose reduction of siRNA Another important clinical phase-1 trial showed high efficacy of an antisense RNA drug targeting apolipoprotein (a), with resulting serum Lp(a) lowering comparable to invasive and expensive lipid apheresis.^137,138

Cardiac-targeted RNA drugs

In contrast to these indirect approaches, the heart may also be targeted directly using recombinant adeno-associated virus (AAV) vectors with high affinity for cardiomyocytes. Selective targeting of cardiofibroblasts, key players in multiple cardiac pathologies,^89–95 is currently not yet possible. Sarcoendoplasmic reticulum Ca²⁺ ATPase (SERCA2a) is the key enzyme for Ca²⁺ uptake during excitation-contraction coupling. Restoration of SERCA2a by gene transfer has proven effective in improving key parameters of HF in pre-clinical studies.³⁷ However, after much promise in early-phase clinical trials, the more recent larger clinical trials have shown disappointing results, so that current vectors, delivery systems, targets, and endpoints have to be reevaluated.³⁷

Beyond these gene therapeutic trials, a search for novel ncRNA targets in HF has been conducted. An assay was developed to functionally screen a whole-genome collection of miRs for selective downregulation of the Ca²⁺ pump.⁸⁶ Through multiple rounds of screening, miR-25 was the most potent miR to elicit a physiological effect comparable to that of siSERCA2a⁸⁶ and was upregulated in myocardial samples from patients with severe HF at the time of cardiac transplantation. miR-25 is expressed primarily in cardiomyocytes of transaortic constriction (TAC)-induced failing mouse hearts, without detectable expression in cardiofibroblasts or vascular endothelial cells, and low expression in vascular smooth muscle. miR-25 inhibition by systemically delivered nanoparticle-coupled antagomir halted HF due to restoration of SERCA2a protein thus increasing SERCA2a supply to undergo post-translational modification e.g. SUMOylation^214,215 resulting in increased enzymatic stability.⁸⁶ Numerous forms of antagomirs were tested, with chemical modifications to confer increased nuclease resistance, higher binding affinity, improved delivery, and modulation of renal clearance. However, Haraguchi et al. proved that a more potent miR inhibition method exists^163,216 which they coined Tough Decoy (TuD). Superior to chemically modified oligonucleotides and Sponge decoys, the TuD RNA is at present the most effective method of miR inhibition^217,218 and, when delivered through a viral vector, confers the longest duration of miR suppression.²¹⁹ AAV9-mediated miR-25 TuD decoy transfer resulted in strong and stable inhibition of cardiac miR-25 levels in HF in vivo (unpublished).

In this review, we do not address other approaches such as the targeting of monocytes-macrophages^{29–31,38,155,220–222} for which we have to refer to other recent reviews.

Non-coding RNAs as biomarkers

Previous sections summarized current knowledge on molecular pathomechanisms and targets from the non-coding genome. Section ‘Non-coding RNAs as biomarkers’ critically reviews ncRNA diagnostics to improve early diagnosis and prognosis assessment. Since ncRNA biomarkers arise from peculiar molecular pathways, they may also help contribute to the identification of patients likely to benefit from specific molecular therapies.

Search for novel biomarkers

Implementation of new biomarkers into clinical practice is an important area of biomedical research. Non-coding RNAs are in particular considered as potential biomarkers if they meet the following criteria: (i) they are quantitatively altered in cardiovascular disease (CVD),^223,224 (ii) they show organ- and cell-specific expression patterns²²⁵ and thus can act as indicators of specific pathogenic processes,¹⁴ (iii) they are easily accessible,²²⁶ and (iv) they withstand conditions such as storage, multiple freeze/thaw cycles and different pHs,²²⁷ show a high degree of stability in body fluids²²⁸ (Box 3).

Non-coding RNA biomarkers in cardiovascular diseases

MicroRNAs

Numerous studies explored the potential of miRs as clinical biomarkers.²²⁴ Focusing on the diagnostic potential, studies compared coronary artery disease (CAD) or AMI patients to controls and assessed the potential of miRs with established biomarkers e.g. troponin.^229–232 miRs and miR-signatures were identified that facilitate differentiation of unstable angina pectoris (UAP) from stable AP (SAP) or non-coronary chest pain (NCCP). These include miRs 134, 198, 370 (UAP/SAP)²³³ and miRs 132, 150, 186 (UAP/NCCP).²³⁴ Similar to research on single miRs, several studies assessed the potential of miRs signatures as biomarkers for CVD. A unique 20-miR signature from whole blood predicted AMI with high specificity and sensitivity, importantly already when troponin was still negative.²³⁵ Out of this signature, a 6-miR signature was subsequently extracted using serial measurements, indicating those miRNAs to be the earliest known markers of AMI.²³⁶

As emphasized in two recent ESC guidelines,^237,238 potentially useful circulatory biomarkers have to undergo state-of-the-art assessment of their added value in CV risk prediction. This is a prerequisite for any new biomarker to gain clinical relevance. Added value of ncRNA markers for risk prediction is of particular interest. Of note, miRs 208b, 519e, 499-5p were strongly associated with increased mortality or HF^239,240 in patients with AMI or DCM, respectively. Zampetaki et al.²⁴¹ identified three miRs that add information to the Framingham Risk Score and may lead to better patient risk stratification. First data from a multicentre prospective study of 1002 STEMI patients suggest that miRs 26b-5p, 320a, and 660-5p discriminated for major cardiovascular events (MACE) within one year of follow-up, and increased risk prediction when added to the GRACE score and a clinical model.²⁸

Other studies evaluating miRs as predictors of disease course, or response to therapy, include viral cardiomyopathies,²⁴² mitral regurgitation before valve repair,²⁴³ or advanced HF before mechanic circulatory support^244–248 (Table 2). As mentioned above, current European guidelines emphasize the unmet need for such biomarkers to help guide clinical decision making in CV prevention and heart failure, and to help transit from usual care to personalized and precision medicine.^238,254

Table 2.

Clinical studies addressing the potential of non-coding RNAs as biomarkers

Selected circulating RNA biomarker studies

Acute myocardial infarction

miR-1, miR-133, miR-499230

miR-208b, miR-133a, miR-400, miR-223250

miR-1, miR-133, miR-208a/b, miR-423, miR-499231

miR-208b, miR-499, miR-320a232

miR signature235,236,241

Acute coronary syndromes

miR-1, miR-133, miR-208b249

miR-133a, miR-499, miR-208251

miR-132, miR-150, miR-186234

miR signature233

MACE prediction after STEMI28

Atherosclerosis

LIPCAR193

aHIF, ANRIL, KCNQ1OT1, MIAT, MALAT1251

CoroMarker253

LncPPARδ254

Long non-coding RNAs

Besides miRs, lncRNAs have increasingly attracted attention in the biomarker field. Kumarswamy et al.¹⁹³ performed transcriptomics analyses of plasma lncRNAs in patients with cardiac remodelling post-AMI and identified LIPCAR (long intergenic non-coding RNA predicting cardiac remodelling) as potential biomarker for HF. Higher levels of LIPCAR were also associated with a higher risk of cardiovascular mortality in HF patients, on top of classical risk stratification, indicating the prognostic potential of LIPCAR. Another study²⁵¹ assessed lncRNAs of AMI patients in whole blood which is deemed as a more reliable lncRNA source.²⁵⁵ lncRNA levels (aHIF, ANRIL, KCNQ1OT1, MIAT, MALAT1) were deregulated, but subsequent assessment of the value of these lncRNAs to predict left ventricular (LV) dysfunction post-AMI showed only weak predictive potential. However, ANRIL and KCNQ1OT1 enhanced LV dysfunction prediction in a multiparameter model. Most recently, lncRNAs CoroMarker²⁵² and LncPPARδ²⁵² were reported as predictive biomarkers for CAD. Despite these progresses, the cellular origin of circulating lncRNAs is mostly unclear and little knowledge regarding causal involvement in the underlying disease is currently available

Circular RNAs

Recently the ncRNAs class of circular RNAs (circRNAs) gained interest as potential biomarkers.^53,256 circRNAs are highly stable in their circular state, abundant, and evolutionarily conserved²⁵⁷ and can be reproducibly detected at high levels in peripheral blood and other body fluids.²⁵⁸ As circRNAs are involved in a wide range of biological processes, deregulation of circRNAs may lead to abnormal cellular functions and diseases. However, their regulation in CVD remains largely unexplored and upcoming studies will have to further explore the potential of circRNAs as clinically relevant biomarkers.

Challenges in using non-coding RNAs as biomarkers

Although ncRNA biomarker research is rapidly advancing, this field still faces several challenges due to preanalytical and analytical factors influencing data quality.²⁵⁹ These factors include the choice of material, sample isolation, detection and processing techniques as well as normalization strategies and the influence of drugs and other, non-cardiac disease and phenotypes. Box 3 provides an overview of technical and clinical aspects to be considered in ncRNA biomarker research. The choice of material (serum, plasma, urine, cells) needs to be considered as alterations in ncRNA levels between these materials have been found. miR levels can greatly vary between serum and plasma,²⁶⁰ and material-dependent stabilities were found for lncRNAs CoroMarker and LncPPARδ between PBMCs and plasma.²⁶¹ In addition, heparin administration prior to blood sampling, haemolysis of blood cells, and intake of medication have been shown to interfere with results of miR quantification.^262,263 Moreover, non-cardiac diseases and phenotypes may influence the circulating levels.^264,265 Methodological issues are another ongoing challenge to cope with: RNA yield greatly varies between different isolation procedures²⁶⁶ and the most striking pitfall of commonly used methods for ncRNA detection (PCR, microarrays, sequencing) is the low correlation between these techniques.²⁶⁷ Furthermore, current normalization strategies to report ncRNA levels are generally not standardized. Different studies use different normalization strategies, ranging from relative normalization with endogenous references (e.g. RNU6, miR-16, let-7a), or exogenous references (e.g. Caenorhabditis elegans miR-39/54, Quanto EC1/2), for calculation of mean values over all miRs and absolute data normalization. Taken together, elimination of ncRNA variability due to technical and analytical factors is of great importance. Reliable isolation methods, cross-platform accuracy, and standardization are needed to generate robust and reproducible results. This is in particularly important to automatize work flows and to polish the way for translation of ncRNA research to clinical routine.

Outlook for translational non-coding RNA research

This review discussed important recent developments and clinical perspectives of ncRNA research, emphasizing key issues associated with translation of basic science insights and preclinical research into clinically valuable novel diagnostics and therapeutics. Regarding diagnostics, elimination of ncRNA variability due to technical and analytical factors, reliable isolation methods, cross-platform accuracy, and standardization are needed to generate robust and reproducible results. This is in particularly important to automatize work flows and to pave the way to clinical routine.

In clinical trials employing ncRNAs as therapeutic targets (miRs) and those employing ncRNAs as therapeutic tools (siRNAs) stringent patient selection^268,269 has emerged as one key factor for success (Box 2, Figure 6). Validation of reported results is rare and therefore large-scale clinical trials are needed to definitely assess the potential of ncRNAs as clinical therapeutic agents. For future clinical applications, precisely defined patient cohorts in whom one pathomechanism is the sole or dominant cause of disease development and progression as well as defined selection criteria and suitable clinical outcome parameters needs to be highly considered. Accordingly, future clinical trials are likely to focus on patients in whom the additional effort of ncRNA therapeutic strategies is likely to results in clinical success.

Figure 6.

Clinical summary figure. From the practical clinical perspective, high-priority diseases for novel RNA therapeutics are obvious. In order to be of potential use, there should be a predominant pathomechanism addressed and no major irreversible damage present. Practical feasibility requires an efficient ncRNA pharmacological tool with very high safety. Two ongoing clinical trials use advanced state-of-the-art liver-targeting technologies.

Supplementary material

Supplementary material is available at European Heart Journal online.

Conflicts of interest: none declared.

Box 1 Goals and challenges of non-coding RNA therapeutics research

Key characteristics of ncRNA-based drugs, clinical goals, and challenges to overcome in translation from preclinical research to clinical applications.

ncRNA, non-coding RNA.

Box 2 Challenges for future translational research

Patient selection, RNA drug delivery systems, prioritization of target diseases, regulatory affairs.

Box 3 Goals and challenges of non-coding RNA biomarker research

Key requirements for ncRNAs biomarkers, expected clinical benefit, and remaining challenges on the way to routine clinical applications.

ncRNA, non-coding RNAs.

Note added in proof

Results of the ORION-1 trial are available online (Ray KK et al. N Engl J Med 2017 May 17, DOI: 10.1056/NEJMoa1615758).