Posttranscriptional regulation of lipid metabolism by non-coding RNAs and RNA binding proteins

Abstract

Alterations in lipoprotein metabolism enhance the risk of cardiometabolic disorders including type-2 diabetes and atherosclerosis, the leading cause of death in Western societies. While the transcriptional regulation of lipid metabolism has been well characterized, recent studies have uncovered the importance of microRNAs (miRNAs), long-non-coding RNAs (lncRNAs) and RNA binding proteins (RBP) in regulating the expression of lipid-related genes at the posttranscriptional level. Work from several groups has identified a number of miRNAs, including miR-33, miR-122 and miR-148a, that play a prominent role in controlling cholesterol homeostasis and lipoprotein metabolism. Importantly, dysregulation of miRNA expression has been associated with dyslipidemia, suggesting that manipulating the expression of these miRNAs could be a useful therapeutic approach to ameliorate cardiovascular disease (CVD). The role of lncRNAs in regulating lipid metabolism has recently emerged and several groups have demonstrated their regulation of lipoprotein metabolism. However, given the high abundance of lncRNAs and the poor-genetic conservation between species, much work will be needed to elucidate the specific role of lncRNAs in controlling lipoprotein metabolism. In this review article, we summarize recent findings in the field and highlight the specific contribution of lncRNAs and RBPs in regulating lipid metabolism.

Introduction

Work done over the last fifteen years has identified numerous classes of non-coding RNA molecules as critical regulators of gene expression [1, 2]. Among these, microRNAs (miRNAs) are the best characterized. Originally identified by the Ambros and Ruvkun laboratories [3, 4], these small non-coding RNAs control the expression of genes associated with numerous biological processes including development and metabolism [1, 2]. miRNAs primarily regulate the expression of genes by directly binding to the 3′ untranslated region of mRNAs [2]. The interaction between miRNAs and mRNA targets is mediated by the RNA-silencing complex (mRISCs) via miRNA/mRNA target sequence complementarity [2]. A single miRNA can regulate the expression of numerous mRNAs, often associated with the same physiological process [2, 5]. Similarly, the expression of a single mRNA can be controlled by several miRNAs, making the regulation of gene expression by miRNAs remarkably complex. Dysregulation of miRNA expression or genetic variants associated with miRNAs or miRNA binding site loci have been associated with cardiometabolic diseases such as obesity, insulin resistance and atherosclerosis [6, 7]. miRNAs have recently emerged as significant regulators of lipid metabolism and promising therapeutic targets for the treatment of cardiovascular diseases [7-10]. Their conservation between species suggests that miRNA mediated regulation of biological pathways is evolutionarily advantageous, however many miRNAs have been shown to be dysregulated under different disease states. For that reason, miRNAs have a unique therapeutic potential, and different approaches have been undertaken to examine this possibility.

While the role of miRNAs in regulation of lipid and glucose metabolism has been a topic of much research over the last few years [7-10], the role of lncRNAs in controlling lipid homeostasis has just started to emerge. lncRNAs are a heterogeneous group of transcribed RNA molecules ranging from 200 to 100,000 nucleotides in length [11, 12]. Depending on their genomic location relative to established protein coding genes, lncRNAs can be classified as long intergenic ncRNAs (lincRNAs), natural antisense transcripts (NATs), enhancer-like ncRNAs (eRNAs), transcribed ultra-conserved regions (T-UCRs) and circular RNAs (circRNAs) [12, 13]. lincRNAs are distinct transcriptional units located in sequence spaces that do not overlap protein-coding genes. NATs are RNA molecules transcribed opposite to the sense DNA strand of annotated transcription units, while eRNAs are short bidirectional products from enhancers that are not processed. T-UCRs are transcripts from genomic regions evolutionarily conserved among mammalian species. CircRNAs are generated from non-colinear splicing of otherwise protein coding exons. Since they form covalently closed continuous loop and do not have 5′ or 3′ ends, cirRNAs are resistant to exonuclease-mediated degradation and are mostly more stable compared to linear RNAs in cells [14]. LncRNAs can regulate gene expression through a variety of mechanisms, including epigenetic modification of DNA, alternative splicing, and post-transcriptional regulation of mRNA stability and translation [15]. Multiple studies have shown that numerous lncRNAs are regulated during development, exhibit cell type-specific expression patterns, localize to specific subcellular compartments, and are associated with physiological functions such as cholesterol metabolism and disease pathogenesis [15].

In addition to miRNAs and lncRNAs, RNA binding proteins (RBPs), especially turnover- and translation-regulatory RBPs, are known to regulate all aspects of mRNA metabolism including processing, transport, translation, and turnover via different RNA interaction motifs generally present in the 3′UTR of the target mRNA [16]. Recent studies have identified a number of RBPs that are associated with the regulation of cellular lipid metabolism. In this review, we summarize the most important and novel roles of miRNAs, lncRNAs and RBPs in regulating cholesterol homeostasis and lipoprotein metabolism.

miRNA Regulation of Cellular Cholesterol Metabolism

miRNA regulation of cholesterol biosynthesis

Intracellular cholesterol levels are tightly regulated by feedback mechanisms controlling the de novo cholesterol biosynthesis, transport, uptake, and efflux [17, 18]. Cholesterol biosynthesis is regulated by a multi-enzyme pathway that includes the 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGCR), the rate-limiting enzyme in cholesterol biosynthesis [19]. Most of the cholesterol biosynthetic enzymes are regulated transcriptionally by the sterol regulatory element-binding protein 2 (SREBP-2) transcription factor [20]. There are three SREBP proteins, among which SREBP-2 primarily activates cholesterol synthesis genes involved in a negative feedback regulation of cellular cholesterol level, whereas SREBP-1a and SREBP-1c have greater effects on genes involved in fatty acid synthesis [20]. Under conditions of high cellular cholesterol, SREBP-2 binds to a SREBP cleavage-activating protein (SCAP)-insulin induced gene (INSIG) complex, leading to a retention of SREBP-2 in the ER. Alternatively, low cellular cholesterol levels result in the disassociation of SCAP from INSIG and promote the translocation of SREBP-2 from the ER to the Golgi [20]. In the Golgi, SREBP-2 is proteolytically cleaved by site-1 protease and site-2 protease. Then, the mature form of SREBP-2 translocates to the nucleus, where it binds to sterol response elements (SREs) present in the promoters of sterol-responsive genes, including HMGCR and the low-density lipoprotein receptor (LDLR), thus promoting cholesterol biosynthesis and uptake [20].

Recent reports have demonstrated that a number of miRNAs regulate cholesterol biosynthesis including miR-223 [21]. miR-223 directly inhibits that expression of 3-hydroxy-3-methyglutaryl-CoA synthase 1 (HMGCS1) and methylsterol monooxygenase 1 (MSMO1), thus suppressing cholesterol biosynthesis [21]. Importantly, mice lacking miR-233 accumulate more free cholesterol in the liver compared to WT mice [21]. In addition to miR-223, other miRNAs including miR-195, miR-21 and miR-29 regulate cholesterol biosynthesis by inhibiting HMGCR expression at the post-transcriptional level [22-24]. miR-195 alters cellular cholesterol and triglyceride levels by directly targeting key enzymes of de novo lipogenesis including HMGCR, acetyl-CoA carboxylase (ACC) and fatty-acid synthase (FASN), resulting in reduced cancer cell proliferation, invasion and migration [24]. miR-21 and miRNA-29 contribute to hepatic free cholesterol accumulation in mouse models of NAFLD and non-alcoholic steatohepatitis by targeting and repressing HMGCR expression [22, 23].

In addition to the direct regulation of HMGCR expression, a number of miRNAs have also been reported to control cholesterol synthesis indirectly by targeting SREBP-2 and INSIG. miR-185 was identified as a regulator of de novo cholesterol biosynthesis and LDL uptake by targeting and repressing SREBP2 expression in hepatic cells [25]. Interestingly, miR-185 expression is regulated by SREBP-1c, suggesting that SREBP-1c and miR-185 might act in concert to maintain cellular lipid homeostasis [25]. miR-122, the most abundant miRNA in the liver, also regulates cholesterol and fatty acid biosynthesis [26, 27]. However, the molecular mechanisms and the direct miR-122 target genes that mediate this effect remain to be elucidated. INSIG1 expression is regulated by miR-92a and miR-145 [28, 29]. The inhibition of INSIG1 expression by miR-92a raised intracellular cholesterol levels and Golgi volume, enhancing protein secretion [28, 29]. Additionally, miR-96 negatively regulates the expression of INSIG2, altering nuclear SREBP levels and endogenous lipid synthesis [30]. The miR-96/182/183 miRNA cluster modulates LDL-C metabolism by regulating SREBP2- mediated transcription of cholesterol biosynthetic enzymes and LDLR. SREBP-2 controls the transcription of the miR-96/182/183 locus, which in turn regulates the expression of key proteins responsible for the maturation and activation of SREBP [30]. miR-96 and miR-182 inhibit the expression of INSIG-2 and Fbxw7 proteins, which control the retention of the SREBP/SCAP complex in the ER and the degradation of nuclear SREBP, respectively [30].

Taken together, these studies suggest that miRNAs regulate de novo cholesterol synthesis by inhibiting the expression of cholesterol biosynthetic enzymes such as HMGCR and by targeting the transcription factors (SREBP-1 and SREBP-2) that control the expression of these enzymes (Figure 1). Additionally, miRNAs also regulate the expression of the molecular machinery (SCAP, INSIG, Fbx7) that controls the processing and stability of these transcription factors. Further experiments will be essential to delineate the specific involvement of these miRNAs in controlling hepatic lipid homeostasis and lipoprotein metabolism in vivo.

Figure 1. miRNA regulation of intracellular cholesterol metabolism.

Schematic overview of miRNAs associated with the regulation of cellular cholesterol homeostasis. Intracellular cholesterol metabolism can be regulated through 3 different ways including cholesterol de novo synthesis, cholesterol transport and esterification. miRNAs repress the cholesterol biosynthesis by directly targeting the rate-limiting enzyme, 3-hydroxy-3-methylglutaryl coenzyme A reductase (HMGCR), or indirectly targeting the important regulators in biosynthesis pathway including sterol regulatory element-binding protein-2 (SREBP-2) and insulin induced gene (INSIG). Intracellular cholesterol transport is hampered by miRNAs targeting cholesterol transporters Neimann–Pick C1 (NPC1) and oxysterol-binding protein-like 6/9 (OSBPL6/9). miR-9 controls cholesterol esterification and lipid droplet formation by directly targeting acylcoenzyme A-cholesterol acyltransferase (ACAT). The indicated miRNAs also regulate the expression of mRNAs associated with cholesterol uptake and export including the low-density lipoprotein receptor (LDLR) and ATP-binding cassette A1 (ABCA1) and G1 (ABCG1) respectively.

miRNAs and cholesterol uptake

While the role of miRNAs in regulating cellular cholesterol efflux has been well characterized, the importance of miRNAs in controlling cholesterol uptake via the LDLR has just recently been elucidated. Of note, work from our group and others has identified miR-27a/b, miR-185, miR-199a, miR-148a, miR-128-1, miR-130b, and miR-301, as direct regulators of LDLR expression and activity in mice and humans [10, 31-35]. Overexpression of these miRNAs markedly reduces LDLR expression levels and attenuates cholesterol uptake in hepatic cells. Most importantly, treatment of mice with antagonists of miR-128-1, miR-148a and miR-185, significantly increases hepatic LDLR expression and reduces circulating LDL-C levels [32, 34, 35]. The role of these miRNAs in regulating LDL-C metabolism and their impact on hepatic lipid metabolism will be discussed in greater detail in a separate section of this review article.

miRNAs and cholesterol efflux

Our group and others originally identified miR-33a and miR-33b as critical regulators of cellular lipid homeostasis and lipoprotein metabolism. miR-33a and miR-33b are encoded within the intronic sequences of the SREBP-2 and SREBP-1 genes. While miR-33a is highly conserved across species, miR-33b expression is lost in small mammals including rodents. Because miR-33a and miR-33b are encoded within the genes of the SREBP transcription factors, multiple groups sought to determine whether these miRNAs might also be involved in regulation of lipid metabolism. Indeed, a number of high impact publications have demonstrated that miR-33a/b are co-transcribed along with their host genes and can function in a synergistic manner. In response to low sterol levels, co-transcription of miR-33a along with SREBP-2 prevents further cholesterol loss by targeting the cholesterol transport molecules ATP-binding cassette transporter A1 (ABCA1) and ATP-binding cassette transporter G1 (ABCG1), while SREBP-2 induces expression of genes involved in cholesterol synthesis and uptake. Similarly, miR-33b and SREBP-1 are co-transcribed, and can work in a concerted fashion through SREBP-1 mediated induction of fatty acid synthesis, and the ability of miR-33 to target genes involved in fatty acid oxidation, including CPT1, HADHB, and CROT. In addition to their ability to regulate intracellular cholesterol levels, miR-33a and miR-33b have been shown to be important regulators of HDL biogenesis and reverse cholesterol transport (RCT), which will be discussed in detail later in this review.

As research into miR-33 as a potential therapeutic target for the treatment of CVD continues to grow, recent reports suggest that miR-33 may also influence cholesterol efflux through additional mechanisms independent of its regulation of ABCA1 and ABCG1 expression. miR-33 has been shown to target AMPK, a nutrient sensitive kinase involved in regulation of key metabolic functions including induction of mitochondrial biogenesis, through the phosphorylation and activation of peroxisome proliferator-activated receptor γ coactivator 1-α (PGC1-α). Interestingly, PGC1-α the master regulator of mitochondrial biogenesis, is also regulated by miR-33. More recently, Karunakaran and colleagues have demonstrated that in addition to PGC1-α, miR-33 can target a number of PGC1-α target genes involved in mitochondrial function, including pyruvate dehydrogenase kinase isozyme 4 (PDK4), and solute carrier family 25 (SLC25A25). This work further showed that alterations in miR-33 can impact mitochondrial respiration and ATP production, which alters ABCA1- mediated cholesterol efflux [36].

miRNAs and intracellular cholesterol transport

Intracellular cholesterol transport is essential for the maintenance of cellular cholesterol homeostasis by delivering cholesterol to the appropriate sites of cholesterol synthesis, esterification, uptake and efflux [17, 18]. Neimann-Pick C1 (NPC1) and NPC2 are two crucial regulators of cholesterol transport in late endosomes and lysosomes [17, 18]. The fatal neurodegenerative disease NPC, an autosomal-recessive lipid storage disorder, is characterized by intracellular free cholesterol accumulation in late endosomes and lysosomes [37, 38]. Native NPC1 is a late endosomal membrane protein including a sterol sensing domain that is also present in important sterol-regulated proteins, such as HMGCR and SCAP, whereas NPC2 presents in the lumen of late endosomes and lysosomes [39]. Cells with mutated forms of NPC1 show massive storage of cholesterol and glycosphingolipids in late endosomes and lysosomes.

NPC1 was firstly reported by our lab as a direct and conserved target gene of miR-33 [40]. Interestingly, the passenger strand of miR-33, miR-33* also targets and represses NPC1 expression as well as other key enzymes involved in cholesterol efflux and fatty acid metabolism [41]. Cells from patients with NPC disease have up to 90% less 25- and 27-hydroxycholesterol in response to LDL loading, indicating that endosomal cholesterol doesn't reach the sites of oxysterol generation. However, more evidence is required to elucidate how targeting of NPC1 by miR-33 and miR-33* impacts intracellular cholesterol transport and regulation of cholesterol homeostasis.

Work by Ouimet and colleagues demonstrated that miR-33 targets the endolysosome and endoplasmic reticulum-associated oxysterol binding protein-like 6 (OSBPL6), which leads to impaired intracellular cholesterol trafficking through abnormal endosome clustering, drop in cholesterol esterification and build up in levels of free cholesterol [42]. The authors found that miR-33 targeting of OSBPL6 is well conserved in mice and non-human primates and hepatic expression of OSBPL6 show positive correlation with circulating levels of high-density lipoprotein cholesterol in humans. These studies further support the potential beneficial role of miR-33 targeted therapies in patients with cardiometabolic diseases.

Defective autophagy in macrophages gives rise to several pathological processes such as impaired cholesterol metabolism and faulty efferocytosis that consecutively contribute to atherosclerosis. Recent reports have demonstrated that miR-33 targets key regulators and effectors of autophagy in macrophages to reduce lipid droplet catabolism, an important process to liberate free cholesterol for efflux [43, 44]. In addition, miR-33 inhibits apoptotic cell clearance via an autophagy-dependent mechanism. Thus miR-33 targeting of autophagy might contribute to its regulation of cholesterol homeostasis and atherogenesis.

miRNA regulation of cholesterol esterification and lipid droplet formation

Excess intracellular cholesterol represses de novo cholesterol biosynthesis and simultaneously activates acylcoenzyme A: cholesterol acyltransferase (ACAT), an ER-localized enzyme that converts cholesterol into cholesterol esters for storage along with triglycerides in the cores of cytosolic lipid droplets [45]. Esterification of free cholesterol mediated by ACAT plays a protective role in macrophages during foam cell formation, avoiding the cytotoxic effects of free cholesterol accumulation in the ER. Taking advantage of a high-content assay performed in human hepatocytes, Whittaker and colleagues identified 11 miRNAs that regulate intracellular lipid droplet formation [46]. Of note, miR-9 directly targets the 3′-UTR of ACAT1 and decreases the formation of foam cells in THP-1-derived macrophages [47]. Additionally, miR-155 is reported to regulate ACAT1 expression indirectly through the NOX2/NFkB pathway in vascular smooth muscle cells (VSMC) [48].

miRNA Regulation of High Density Lipoprotein Metabolism

Abnormal levels of cholesterol are deemed to be a major risk factor for cardiometabolic diseases, such as type-II diabetes and atherosclerosis [49, 50]. Therefore, it is necessary to explore potential therapeutic targets to retard cholesterol accumulation and promote cholesterol removal. Reverse cholesterol transport (RCT) is an important biological process to reduce excess cholesterol and transport from the peripheral tissues to the liver, where cholesterol can be reutilized or excreted into feces [51, 52]. Lipid-poor HDL and its major protein component, apolipoprotein AI (apoAI), accelerate cholesterol movement from peripheral cells to the liver via ATP-binding cassette transporters ABCA1 and ABCG1 [51, 52]. Over the last decade, multiple of studies have demonstrated the pivotal role of miRNAs in regulating HDL-C metabolism [9].

miR-33, a major regulator of HDL-C metabolism

Among the miRNAs involved in regulation of HDL-C metabolism, miR-33 is deemed to be one of the most important, due to its ability to regulate numerous key functions including cholesterol efflux, transport and uptake; HDL biogenesis; bile acid synthesis and secretion; and fatty acid degradation [40, 53-58]. In addition to its role in facilitating cellular cholesterol efflux, ABCA1 in the liver is critical for the biogenesis of HDL, the primary acceptor of cholesterol following efflux from peripheral tissues [59, 60]. As such, multiple studies have shown that inhibition or genetic deletion of miR-33 increases hepatic ABCA1 expression and circulating HDL-C in mice and non-human primates [40, 56, 57, 61, 62]. Most importantly, antagonism of miR-33 in mice attenuates the progression and enhances the regression of atherosclerotic plaques [63, 64]. However, other studies with miR-33 inhibitors have not observed any differences in circulating HDL-C levels or atherosclerotic plaque development [65]. The reason behind these discrepancies could be explained by differences in the antisense oligonucleotide (ASO) chemistry, duration of the treatment and type of diet used in these studies. Interestingly, the three reports that found significant reduction in atherosclerotic burden without affecting circulating HDL-C also demonstrated the efficacy of anti-miR-33 therapy in targeting atherosclerotic plaque macrophages and the whole aorta [36, 63, 64, 66, 67]. These findings correlate with a recent study published from our group demonstrating that genetic ablation of miR-33 in hematopoietic cells protects against the progression of atherosclerosis without altering plasma HDL-C levels [68].

Additionally, silencing of miR-33 has been shown to promote bile acid secretion by increasing the expression of ATP8B1 and ABCB11 in vivo [53]. Moreover, miR-33 knockdown also elevated CYP7A1 levels, thereby enhancing bile acid synthesis [69]. Based on the above findings, miR-33 is a critical regulator of cholesterol homeostasis due to its ability to promote cholesterol efflux (via ABCA1 and ABCG1), enhance HDL biogenesis (via ABCA1), and increase bile acid synthesis (via CYP7A1) and secretion (via ATP8B1 and ABCB11).

Other miRNAs and HDL-C metabolism

In addition to miR-33, other miRNAs control HDL-C metabolism by targeting ABCA1 [9]. Inhibition of miR-148a and miR-144 increases hepatic ABCA1 expression and plasma HDL-C levels in mice [32, 34, 70, 71]. miR-148a is highly abundant in monocytes and macrophages, and its inhibition reduces macrophage foam cell formation by promoting cholesterol efflux [34]. Similarly, miR-302a was found to target ABCA1, and anti-miR-302a treated mice displayed reduced atherosclerotic plaque size [72]. Scavenger Receptor class B member 1 (SR-B1) is a member of the CD36 superfamily involved in HDL-C metabolism, cholesterol clearance and transport [73]. SR-B1 positively regulates RCT by enhancing clearance of cholesterol to the feces. Several studies have uncovered miRNAs, including miR-96, miR-185, miR-455, miR-125a and miR-223, that bind directly to the 3′ UTR of SR-B1 and inhibit its expression, and inhibition of these miRNAs was shown to increase HDL-C clearance [21, 35, 74].

The liver X receptor (LXR) is a central transcriptional regulator of cholesterol metabolism by targeting multiple genes including ABCA1, ABCG1, inducible degrader of the LDLR (IDOL), SREBP-1c, fatty acid synthase (FASN), phospholipid transfer protein (PLTP) and ADP-ribosylation factor (ARF)-like 7 (ARL7). LXR activation is capable of inhibiting macrophage-driven inflammation, enhancing ABCA1-dependent RCT, and promoting an increase in serum HDL. Recently, miR-155, which is elevated in the peripheral blood of patients with non-alcoholic fatty liver disease, was found to regulate SREBP-1 and FASN by directly targeting LXR-α [75]. miR-155 inhibitors markedly increased the level of LXR-α, thereby enhancing intracellular lipid content [75]. Similarly, miR-206 and miR-1 directly repressed LXR-α expression, resulting in decreased lipogenesis, reduced lipid droplet accumulation and increased cholesterol efflux in hepatocytes and macrophages [76]. Therefore, it may be possible to regulate LXR levels through miRNA based therapies, and thereby influence numerous key factors involved in regulation of cholesterol homeostasis.

HDL as a vehicle for the transport and delivery of miRNAs

As the primary acceptor of cholesterol effluxed from peripheral tissues, HDL-C is a key player in the global regulation of cellular cholesterol homeostasis. Surprisingly, work has revealed that miRNAs, such as miR-375 and miR-223 can also be found in HDL particles, indicating a previously unexplored role for HDL, and perhaps other lipoproteins, in transporting regulatory RNAs [77]. Vickers et al. demonstrated that HDL could increase intracellular miR-105-5p in familial hypercholesterolemia patients, while there was no obvious change in healthy volunteers [77]. In a follow up study, Tabet and colleagues, demonstrated for the first time that HDL-transfer of miRNAs to endothelial cells (ECs) can regulate EC activation [78]. These observations suggest that the anti-inflammatory effect mediated by HDL in ECs might be conferred, in part, by the repression of ICAM-1 expression via HDL-transfer of mi-223. Although work by several groups has sought to further illuminate the relationship between miRNAs and HDL metabolism, additional studies are still needed to address critical questions, including the mechanism of miRNA loading in HDL particles and the receptors that mediate the transfer of HDL-derived miRNAs to cell types relevant to CVD (macrophages, ECs, and VSMCs).

miRNA Regulation oF VLDL-C and LDL-C Metabolism

While the role of miRNAs in regulating HDL-C metabolism has been studied extensively, the importance of miRNAs in controlling circulating LDL-C is still poorly understood. Nevertheless, a number of studies have recently reported the importance of several miRNAs that control circulating LDL-C levels by modulating very low-density lipoprotein (VLDL) secretion and LDL clearance [10] (Figure 2).

Figure 2. Regulatory role of miRNAs in LDL-Cholesterol metabolism.

Schematic representation of miRNAs that regulate LDL-C metabolism. VLDL indicates very low-density protein; LDLR, low-density lipoprotein receptor; NSF, N-ethylmaleimide-sensitive factor; MTTP, microsomal triglyceride transfer protein; FA, Fatty acid. Figure was created using the Servier Medical Art illustration resources (http://www.servier.com).

miRNAs that modulate VLDL secretion

Perturbations in circulating levels of ApoB-containing lipoproteins such as VLDL and LDL have a marked effect in the development of atherosclerosis [49, 50]. Early studies have uncovered a number of miRNAs that regulate plasma VLDL-C and LDL-C levels in mice and non-human primates.

miR-122 and miR-30c in lipid metabolism and atherosclerosis

The first miRNAs identified as important regulators of circulating LDL-C in vivo by controlling VLDL secretion were miR-122 and miR-30c [26, 27, 79, 80]. While miR-122 is the most abundant miRNA in the liver, accounting for about 70% of the total miRNA content, miR-30c levels are very low. Importantly, genetic ablation or pharmacological inhibition of miR-122 in mice and non-human primates using ASOs results in a marked reduction of total plasma cholesterol and triglyceride (TAG) levels [81, 82]. These effects were further confirmed in mice lacking miR-122 expression. It is thought that the reduced circulating cholesterol observed in mice treated with miR-122 antagonists is due to alterations in a combination of different metabolic processes, including the inhibition of cholesterol and fatty acid biosynthesis, enhanced fatty acid oxidation and decreased VLDL secretion [26, 81, 82]. Despite these intriguing findings, the molecular mechanisms by which miR-122 controls lipoprotein metabolism remain largely enigmatic. Furthermore, while the promising results of animal studies suggest inhibition of miR-122 may be a useful approach for the treatment of CVD, it has also been shown to be a tumor suppressor and inhibition and genetic ablation of miR-122 were found to increase the prevalence of hepatocellular carcinoma [81, 82].

miR-30c also controls circulating cholesterol by decreasing lipid synthesis and hepatic VLDL secretion through direct targeting of the microsomal triglyceride transfer protein (MTTP), an enzyme required for VLDL assembly [79, 80]. Moreover, miR-30c inhibits hepatic lipid synthesis by targeting lysophosphatidyl glycerol acyltransferase 1 (LPGAT1), an enzyme involved in phospholipid synthesis. Consequently, overexpression of miR-30c in mice, using lentiviral constructs or miRNA mimics, significantly reduced plasma cholesterol levels and attenuated the progression of atherosclerosis [79, 80]. Importantly, inhibition of MTTP expression and VLDL production by miR-30c does not impact the accumulation of lipids in the liver, suggesting that therapeutic approaches to elevate miR-30c might be used to manage cholesterol disorders.

miR-33

In addition to miR-122 and miR-30c, several reports have shown that miR-33 might also modulate VLDL metabolism in mice and non-human primates [61, 83]. However, findings are not consistent across studies, yielding confusing and contradictory results. A number of studies in mouse models have shown that inhibition of miR-33 using locked nucleic acid (LNA)- and methoxy-ethyl (MOE)-modified oligonucleotides and lentivirus increases circulating HDL-C without affecting plasma VLDL-C levels [40, 57, 58, 84]. These results are in contradiction to reports that have demonstrated that chronic anti-miR-33 therapy increases circulating TAG levels and hepatic lipid content [83, 85]. Mechanistically, one of these studies identified N-ethylmaleimide-sensitive factor (NSF), an ATPase enzyme involved in intracellular trafficking, membrane fusion and vesicle secretion, as a miR-33 target gene [83]. Additionally, Rayner and colleagues have recently reported that long-term miR-33 inhibition using miR-33 ASOs promotes fatty acid oxidation without affecting body weight or circulating TAGs [84]. The different results obtained in this work compared with a previous study from the Fernández-Hernando lab cannot be explained by differences in diets or the ASO chemistry, which were similar in both studies.

While the studies using miR-33 inhibitors in vivo have shown some discrepancies, two independent groups have demonstrated that genetic loss of miR-33 results in obesity, insulin resistance and dyslipidemia. In the first report, Horie and colleagues showed that miR-33^-/- mice develop obesity and metabolic dysfunction and have increased total cholesterol levels, especially in mice on a high fat diet (HFD), while circulating TAGs were not found to be altered [86]. Unfortunately, these measurements were not performed in earlier work characterizing atherosclerotic plaque formation in mice deficient in both miR-33 and Apoe [87]. More recent work demonstrates mice lacking miR-33 in the Ldlr^-/- mouse model of atherosclerosis also had increased body weight and insulin resistance as well as elevated circulating levels of both cholesterol and TAGs [68]. These pro-atherogenic effects likely offset the beneficial effects observed in mice transplanted with miR-33 deficient bone marrow, as no differences in atherosclerotic plaque formation were observed in whole body miR-33^-/-Ldlr^-/- mice [68]. The molecular mechanism by which loss of miR-33 influences circulating cholesterol and TAGs is not entirely understood and could be mediated by enhanced VLDL production or decreased triglyceride-rich lipoprotein (TRL) clearance. In this regard, absence of miR-33 might enhance the VLDL secretory pathway by de-repressing NSF or could promote VLDL production indirectly by enhancing hepatic insulin resistance, which has been associated with insulin-mediated ApoB degradation [88]. Further studies using tissue specific miR-33 knockout mouse models will be important to dissect the contribution of miR-33 in different metabolic tissues in regulating glucose and lipid metabolism.

The results found in non-human primates are also inconsistent. While one study observed a significant reduction in plasma VLDL TAG and circulating TAG levels [61], the other report did not find differences in plasma TAG levels [62]. While it is not clear why these discrepancies exist between different studies, a number of factors including diet, genetic background, method of miR-33 inhibition and experimental design could contribute to these differences. Indeed, while one study showed a significant increase in hepatic miR-33b levels in response to HFD [61], another report did not find differences in miR-33b expression in the liver in response to high cholesterol diet (HCD) [62]. As such, further studies will be important to clarify the contribution of miR-33 in regulating VLDL production and circulating TAGs.

miRNAs that modulate hepatic LDLR expression and LDL-C clearance

Re-uptake of LDL-C into the liver via the LDLR is critical for proper maintenance of circulating LDL-C levels, and recent work has demonstrated that miRNAs can play an important role in the post -transcriptional regulation of LDLR expression [32, 34]. Specifically, the LDLR was found to be a direct target of numerous miRNAs, including miR-27a/b, miR-128-1, miR-130b, miR-148a, miR-185, miR-199a, and miR-301, which were able to regulate LDLR expression in both mouse and human hepatic cells [32, 34]. Among these miRNAs, only miR-128-1, miR-148a and miR-185 were shown to considerably alter plasma LDL-C in vivo [25, 32, 34, 35]. Here we will discuss how miRNAs contribute to the regulation of hepatic lipid metabolism by modulating circulating LDL-C levels.

miR-148a

Two independent studies by Goedeke et al and Wagschal et al have recently identified miR-148 as an important regulator of hepatic LDLR expression and lipoprotein metabolism both in vitro and in vivo [32, 34]. Through a high-throughput genome-wide screening assay, Goedeke and colleagues identified miRNA-148a as a negative regulator of LDLR expression and activity in human hepatic cells. miR-148a is highly expressed in mouse and human hepatic tissue and conserved among vertebrate species [32]. Its expression is also regulated by dietary lipids and SREBP-1. Interestingly, pharmacological inhibition of miR-148a using ASOs also lowered plasma LDL-C levels in two different mouse models of hypercholesterolemia [32]. This work was corroborated by recent reports that reported SNPs (rs4722551, rs4719841, and rs6951827) present in the promoter region of miR-148a, which are associated with altered plasma cholesterol, LDL-C and TAG levels [34]. Furthermore, a miR-eQTL analysis performed in human livers discovered a strong connection between SNP status and miR-148a expression levels [34]. The precise mechanism by which these SNPs lead to altered plasma lipids remains unknown. These genetic variations might influence the regulation of miR-148a expression via SREBP-1, however this explanation needs additional investigation. The role of miR-148a in modulating lipid metabolism is likely to be more complex and not only explained by its direct targeting of LDLR. In particular, miR-148a was shown to directly target the 3′UTR of other genes involved in lipid metabolism, including PGC1a, AMPK, and INSIG1 [34]. Furthermore, miR-148a also directly targets ABCA1, and miR-148a silencing in vivo increases hepatic ABCA1 expression and circulating HDL-C levels [32, 34]. Thus, these studies emphasize the potential of anti-miR-148a therapies as a treatment strategy for both lowering circulating plasma LDL-C and increasing circulating HDL-C in patients suffering from cardiovascular disease.

miR-128-1

Wagschal et al have recently demonstrated a strong correlation between a number of SNPs in the miR-128-1 gene locus and altered plasma lipid levels [34]. Similar to miR-148a, miR-128-1 directly targets the 3′UTR of the LDLR and ABCA1, and its inhibition in mice caused a significant drop in circulating cholesterol and TAG levels [34]. Furthermore, inhibition of miR-128-1 in vivo improved glucose tolerance and insulin sensitivity. Importantly, miR-128 ASOs enhanced the expression of the insulin receptor (INSR) and insulin receptor substrate 1 (IRS-1), thus increasing insulin sensitivity [34]. Moreover, miR-128-1 controls the expression of genes involved in the fatty acid synthesis, including fatty acid synthase (FASN) and Sirtuin 1 (SIRT1). SIRT1 is an NAD⁺-dependent deacetylase that has been shown to deacetylate a number of key metabolic regulators, including SREBP-1, leading to impaired SREBP-1-dependent lipogenesis [34]. SIRT1 can also increase the activity of PGC1α but through direct deacetylation and by targeting LKB1, a kinase upstream of AMPK [89]. Taken together, these studies indicate that miR-128a expression might affect plasma lipid levels by regulating the expression of genes associated with lipid and glucose metabolism. Additional studies will be essential to more clearly define the impact of miR-128a on target genes in different tissues and their involvement in the maintenance of lipid and glucose homeostasis.

miR-185

Several recent studies have demonstrated the role of miR-185 in regulation of cholesterol metabolism [25, 35]. Various in vivo and in vitro studies have shown that miR-185 can regulate the expression of both LDLR and the hepatic HDL-C receptor SRBI, which facilitates the uptake of cholesteryl esters from HDL in the liver [25, 35]. Remarkably, miR-185 also targets an RNA-binding KH-type splicing regulatory protein (KSRP), that negatively modulates the expression of the human LDLR [25]. Interestingly, inhibition of miR-185 in Apoe^-/- mice reduced plasma cholesterol levels and mitigated the progression of atherosclerosis.

LncRNAs and Cholesterol Metabolism

While the importance of miRNAs for the regulation of cholesterol metabolism has been well established, and ongoing work continues to clarify the mechanisms underlying these regulatory networks, even more recent work has identified another new category of regulatory molecules collectively known as lncRNAs. While work in this field is still in an early stage, lncRNAs, including lincRNAs, NATs, eRNAs, and T-UCRs, have been shown to have dramatic effects on a number of biologic processes, including regulation of lipid metabolism (Figure 3).

Figure 3. Regulation of cholesterol metabolism by lncRNAs and RBPs.

Schematic overview of lncRNAs and RBPs involved in the regulation of hepatic and macrophages cholesterol homeostasis and lipoprotein metabolism. Blue boxes highlight lncRNAs, which regulate genes that control cholesterol metabolism. Orange boxes highlight RBPs that regulate cholesterol metabolism and inflammation in macrophages. FXR indicates farnesoid X receptor; LeXis, liver-expressed LXR-induced sequence; LncLSTR, liver specific triglyceride regulator; ApoA1-AS, ApoA1 antisense; LPL, lipoprotein lipase and TDP-43, TAR DNA-binding protein 43; GRP119, G-protein coupled receptor 119; ZFP36, zinc finger protein 36 homolog; HUR, human antigen R. Figure was created using the Servier Medical Art illustration resources (http://www.servier.com).

LeXis

Tontonoz's group has recently identified LeXis (liver-expressed LXR-induced sequence), a non-coding RNA that lies in close proximity to the canonical LXR target gene Abca1 in mice and is markedly induced in response to LXR agonists and HFD [90]. Hepatic LeXis overexpression reduces circulating cholesterol, attenuates cholesterol biosynthesis and inhibits the expression of cholesterol biosynthetic genes. Conversely, genetic ablation of LeXis or acute pharmacological inhibition using ASOs enhances hepatic expression of genes associated with cholesterol biosynthesis, leading to a significant accumulation of cholesterol in the liver. Mechanistically, the authors found that LeXis interacts with and influences the binding of RALY, a heterogenous ribonucleoprotein that acts as a transcriptional cofactor for cholesterol biosynthetic genes in the mouse liver, to DNA. As a consequence, the actions of LeXis in vivo were dependent on RALY as the ability of LeXis to alter serum cholesterol levels and hepatic gene expression was impaired in the setting of RALY knockdown. Batch coordinate conversion between mouse and human assemblies revealed moderate conservation of the LeXis genomic sequence in a region adjacent to the human ABCA1 gene. However, further experiments are needed to determine whether the putative lncRNA annotated in this region (TCONS_00016452) regulates cholesterol metabolism in humans.

ApoA1-AS and lnc-HC

NAT ApoA1-AS is encoded in the apolipoprotein gene cluster that contains four different transcripts including ApoA1, ApoA4, ApoA5 and ApoC3. ApoA1-AS controls the expression of this apolipoprotein gene cluster epigenetically by recruiting histone-modifying enzymes [91]. Targeting ApoA1-AS using ASOs increases ApoA1 expression in both monkey and human cells and enhances hepatic RNA and protein expression in African green monkeys. While these results are of interest, it is still not known whether the increase in circulating ApoA1 influences plasma lipid levels and/or lipoprotein metabolism. Another lncRNA lnc-HC has been recently identified and was found to be highly expressed in liver [92]. Lnc-HC interacts with hnRNPA2B1 forming a RNA-protein complex, which can then bind to target mRNAs, including Cyp7a1 and Abca1 [92]. Inhibition of Lnc-HC increased CYP7a1 and ABCA1 expression in hepatocytes, thus promoting cholesterol catabolism. Lnc-HC is conserved in humans and rodents and appears to be highly expressed in the liver and adipose tissue [92]. These observations suggest that Lnc-HC might play a role in regulating lipid metabolism. However, additional studies in vivo are needed to define the role of Lnc-HC in regulating hepatic lipid homeostasis and lipoprotein metabolism.

LncLSTR

Liver-specific triglyceride regulator (LncLSTR) was identified using an unbiased screen designed to identify LncRNAs highly expressed in the liver [93]. Of note, specific inhibition of LncLSTR led to a marked reduction in circulating TAGs. Mechanistically, LncLSTR depletion increased apoC2 levels, an activator of the lipoprotein lipase (LPL), thus enhancing VLDL and chylomicron catabolism, leading to an increase in plasma TAG clearance [93]. Hepatic LncLSTR expression is regulated by FXR and forms a complex with TDP-43, thus regulating the expression of Cyp8b1, a critical enzyme involved in bile acid synthesis.

RNA-RNCR3

Shan et al reported that lncRNA-RNCR3 is expressed in atherosclerotic plaques, particularly in ECs and VSMCs, where its expression is enhanced by ox-LDL in vitro [94]. Knockdown of RNCR3 using shRNAs accelerated the progression of atherosclerosis, aggravated hypercholesterolemia and decreased the proliferation of ECs and VSMCs in Apoe^-/- mice. Mechanistically the authors propose that RNCR3 acts as a competing endogenous RNA (ceRNA), and forms a feedback loop with Kruppel-like factor 2 (KLF2) and miR-185-5p to regulate cell function [94]. However, the mechanism of why RNCR3 depletion increases both cholesterol and TAG levels needs to be further explored, and these findings need to be validated using genetic mouse models.

ANRIL

Early studies by Holdt and colleagues found that the expression of antisense noncoding RNA in the INK4 locus (ANRIL) in the plasma and atherosclerotic plaques of human patients directly correlated with the severity of atherosclerosis [95]. The linear form of ANRIL mediates atherosclerosis risk through trans-regulation of gene networks leading to pro-atherogenic cellular properties, such as increased proliferation and adhesion, which depends on the ALU motifs within ANRIL RNA [96]. Recent work by the same authors demonstrates that circular form of ANRIL (circANRIL) regulates the maturation of precursor ribosomal RNA (pre-rRNA), thus controlling ribosome biogenesis and nucleolar stress. circANRIL confers disease protection by modulating apoptosis and proliferation in human vascular cells and tissues, which are key cellular processes governing atherogenesis [97].

LncHR1

Li et al have recently identified a novel human specific lncRNA, lncHR1, as a negative regulator of SREBP-1c expression [98]. Overexpression of lncHR1 inhibited expression of SREBP-1c and FASN leading to decreased oleic acid-induced hepatic cell TAG and lipid droplet (LD) accumulation. Importantly, overexpression of ncHR1 in transgenic mice decreased hepatic expression of SREBP-1c, FASN and Acetyl-CoA carboxylase α (ACCα), and reduced hepatic and plasma TAGs in mice fed a high-fat diet [98]. The exact mechanism by which lncHR1 controls SREBP-1c expression and activity is not clear and needs further research.

DYNLRB2

DYNLRB2 was recently identified as one of the differentially regulated lincRNAs in macrophage foam cells [99]. Authors showed that loading with oxidized-LDL (ox-LDL) increased the expression of DYNLRB2 in human THP1 macrophages. Overexpression of DYNLRB2 in THP1 cells resulted in a decrease in the level of TAGs, free cholesterol, and cholesteryl esters through up-regulation of the G protein-coupled receptor 119 (GPR119) and ABCA1, and the subsequent increase in cholesterol efflux to ApoA1 molecules. Furthermore, loss of GPR119 in mice led to a reduction of circulating lipids and the development of atherosclerosis suggesting loss of DYNLRB2 would produce similar results [99]. However, further investigation is necessary to assess the direct physiological effects of DYNLRB2.

RNA-Binding Proteins and Cholesterol Metabolism

In addition to lncRNAs, RNA binding proteins (RBPs) are known to regulate all aspects of mRNA metabolism including processing, transport, translation, and turnover via different RNA interaction motifs generally present in the 3′UTR of the target mRNA [16]. Recent studies have identified number of RBPs that are associated with cholesterol and lipid metabolism and cardiovascular disorders. Human antigen R (HuR), also known as embryonic lethal, abnormal vision Drosophila-like 1 (ELAVL1), is a ubiquitously expressed RBP belonging to the Hu/Elav family, that modulates the stability and translational efficiency of mRNAs by interacting with uridylate (U)-rich or adenylate-uridylate (AU)-rich elements in the 3′UTR of target genes [100]. Ramirez and colleagues demonstrated that HuR interacts directly with ABCA1 mRNA, increasing ABCA1 protein expression and promoting cellular cholesterol efflux. Interestingly, this interaction was enhanced by cellular cholesterol, which increased the mRNA and protein levels of HuR. Conversely, statins attenuate HuR translocation from the nucleus to the cytoplasm and impair HuR-mediated stabilization of ABCA1 mRNA. Of note, HuR is highly expressed in macrophages accumulated in mouse and human atherosclerotic plaques, suggesting an important role for HuR in controlling macrophage lipid homeostasis in vivo [101].

RBPs can function in a context dependent manner to regulate a specific set of RNAs. Zinc finger protein-36 (ZFP36) is minimally expressed in the healthy aorta but is highly expressed in endothelial cells overlying atherosclerotic lesions and in foam cells [102]. Its expression is induced as a protective mechanism against inflammation, reducing the expression of inflammatory cytokines including IL-6 and MCP-1 by inhibiting the transcriptional activation of NF-κB, as well as by binding to AU rich elements of cytokine mRNAs and decreasing their transcript stability [102]. However, the physiologic role of ZFP36 during the progression of atherosclerosis requires additional investigation.

Recently, Mobin and colleagues identified the RBP VIGILIN as a novel regulator of hepatic VLDL secretion. VIGILIN binds to the CU-rich regions of ApoB and other pro-atherogenic secreted proteins including apolipoproteinC-III and fibronectin, promoting their translation. Consequently, hepatic inhibition of VIGILIN decreases hepatic VLDL secretion and circulating LDL-C levels, leading to a significant protection against the progression of atherosclerosis in Ldlr ^-/- mice [103]. Interestingly,

VIGILIN expression is elevated in the livers of obese mice and patients with fatty liver disease [103], suggesting that VIGILIN could contribute to the increased hepatic VLDL production observed in obese and insulin resistance subjects.

Concluding Remarks

Work in recent years has clearly established the important contributions of novel classes of RNA regulatory molecules, namely miRNAs and lncRNAs, and RBPs in the regulation of lipid metabolism. As the most well-established of these, miRNAs have been demonstrated to impact many different facets of lipid metabolism. Moreover, the ability of an individual miRNA to target multiple mRNAs allows for more complex levels of regulation, such as the ability of miR-148 to target both the LDLR and ABCA1, resulting in increased LDL-C and decreased HDL-C, respectively. However, this same ability has raised serious questions about potential unintended impacts of miRNA based therapeutic approaches. The increased prevalence of hepatocellular carcinoma following anti-miR-122 treatment and the dramatic obesity phenotype of miR-33 deficient mice on a HFD highlight the validity of these concerns. As such, new tools will be needed to better understand the specific mechanisms by which miRNAs mediate their effects and assess the overall impact of miRNA base therapies. Of note, the dramatic obesity and insulin resistant phenotype observed in mice lacking miR-33 is remarkable compared to mice treated with miR-33 ASOs, suggesting that miR-33 might control whole body metabolism by acting in an organ that is not targeted efficiently by miR-33 inhibitors.

More recent findings have demonstrated the strong impact lncRNAs and RBPs can have on regulation of cholesterol metabolism, and since the human genome encodes thousands of lncRNAs, it is expected that in the near future other lncRNAs will also be identified. Despite this exciting future, the analysis and the identification of functional lncRNAs will be challenging because of the modest conservation of these RNA molecules between species. Additionally, a more complete understanding of the mechanisms by which these lncRNAs exert their effects, and additional studies directly assessing the impact of the lncRNAs on atherosclerotic plaque formation will be needed to properly assess the therapeutic potential of lncRNA based approaches. RBPs provide an additional post-transcriptional regulatory mechanism to modulate the levels of proteins involved in cholesterol and lipid homeostasis and disease development. However, our current understanding of these proteins is still limited and requires further investigation to elucidate the direct and indirect mechanisms through which they may regulate target mRNA expression, stability, and translation.

Overall, work over the past fifteen years has identified multiple new levels of regulatory networks, and the discovery of miRNAs and lncRNAs has entirely transformed the way we view the genetic code. In this review article, we summarize recent work demonstrating the important role of these regulatory molecules in the control of lipid metabolism. While much work is still needed to properly understand the regulation, impact, and mechanistic functions of these new modulators of gene regulation current work has established the modulation of these non-traditional regulatory molecules has strong potential for the development of novel therapeutic approaches for the treatment of cardiovascular disease.