Nuclear Receptors and microRNA-144 Coordinately Regulate Cholesterol Efflux

Abstract

The ATP-binding cassette transporter A1 (ABCA1) is a key mediator of cellular cholesterol efflux and HDL maturation. ABCA1 mRNA has an unusually long 3’ untranslated region, which makes it highly susceptible to microRNA (miRNA) targeting and repression. As such, multiple miRNAs have been reported to directly target ABCA1, including miR-33a/b, miR-26, miR-106b, and miR-758. Many of these miRNAs participate in feed-forward or feedback networks in controlling cholesterol and lipid homeostasis. Antisense oligonucleotide-based inhibition of miR-33 was found to increase HDL-C levels and regress atherosclerosis in mice and non-human primates. In this edition of Circulation Research, two separate studies identified novel miRNA networks driven by nuclear receptor induced miR-144 targeting of ABCA1 and cholesterol efflux. The first study reports that miR-144 serves to buffer uncontrolled ABCA1 activation in response to high cholesterol states and liver X receptor (LXR) activation in macrophages and liver. The second study highlights the role of miR-144 and ABCA1 in promotion of bile acid secretion in response to farensoid X receptor (FXR) activation in the liver. These studies suggest that anti-miR-144, like anti-miR-33, could be a novel approach to targeting HDL and reverse cholesterol transport.

Proper cholesterol homeostasis requires a complex network of sterol-sensing proteins, membrane dynamics, and extensive regulation of transcription, translation, post-translation modifications, and protein turnover^1-3. Together, multi-layered regulatory modules control three key processes to balance cellular cholesterol levels – de novo cholesterol biosynthesis, cholesterol uptake though lipoprotein receptors, and cholesterol efflux or excretion into bile^{1, 4, 5}. In addition to maintaining homeostasis, cholesterol efflux to lipid-poor apolipoprotein A-I (apoA-I) serves to form high-density lipoproteins (HDL) particles, which provide systemic cholesterol homeostasis through the reverse cholesterol transport pathway⁶. The ATP-binding cassette transporter A1 (ABCA1) is a critical transporter of cholesterol and lipids from cells to extracellular apoA-I, a process that protects against cholesterol overload. ABCA1 expression is regulated by key nuclear receptors, namely the liver X receptor (LXR) family and their heterodimeric partners, retinoic acid receptors (RXR)^7-9. As cellular sterol levels increase, cholesterol becomes oxygenated and accumulating oxysterols activate LXR/RXR to drive ABCA1 expression, and thus, cholesterol efflux from the cell^{3, 10}.

In the liver, ABCA1-mediated cholesterol efflux to newly synthesized lipid-poor apoA-I accounts for the formation of the majority of nascent HDL particles; therefore, hepatic regulation of ABCA1 is an important regulator of plasma HDL cholesterol (HDL-C) levels and systemic cholesterol homeostasis¹¹. Catabolism of HDL-derived cholesterol also occurs in the liver through conversion excretion into the bile as free cholesterol or bile acids. The Farnesoid X receptor (FXR) is a nuclear receptor expressed in the liver and intestine that controls hepatic sterol and bile acid levels through transcriptional regulation of bile acid and lipid-associated genes¹². Similar to LXR, increased cellular cholesterol levels and bile acid accumulation activate FXR to promote bile acid secretion.

miRNAs are small non-coding regulatory RNAs that bind to complementary sites within mRNA 3’ untranslated (3’ UTR) and coding regions and provide post-transcriptional gene regulation through translation inhibition and mRNA degradation^13-16. Often viewed as biological rheostats or buffers of gene expression, the functional relevance of miRNAs in lipid homeostasis is striking, and dysregulation of miRNA-mediated mechanisms are now viewed as etiological causes of metabolic disease. Although multiple miRNAs have been found to regulate lipid metabolism, namely miR-27b¹⁷ and miR-122¹⁸, miR-33a/b has been extensively studied and represents perhaps the strongest rationale for miRNAs as key mediators of cholesterol homeostasis. miR-33a/b is co-transcribed from introns within sterol regulatory element-binding transcription factor 1 (SREBF1, miR-33b) and SREBF2 (miR-33a)^19-21. During low sterol conditions, SREBF transcription factors and miR-33a/b are co-transcriptionally activated to, respectively, increase cholesterol biosynthesis and reduce cholesterol efflux through repression of ABCA1 expression. miR-33a/b directly targets ABCA1 mRNA, which harbors 4 putative miR-33 target sites in its 3’ UTR. In addition to ABCA1, miR-33a/b has also been found to target many other cholesterol, lipid, and bile acid-associated genes, including Niemann-Pick C1 (NPC1), phospholipid-transporting ATPase IC (ATP8B1), ATP-binding cassette transporter B11 (ABCB11), and ATP-binding cassette transporter G1 (ABGC1)^21-24. Importantly, inhibition of miR-33 in mice and non-human primates was found to increase HDL-C levels and in mice to reduce atherosclerosis^{19, 20} and this approach is under development as a novel therapy for atherosclerotic cardiovascular disease²⁵.

miRNAs recognize mRNA targets though seed-based complementarity, 5’ bases 2-8 of the mature miRNA¹⁵; therefore, one miRNA has the potential to target many mRNAs, and one gene (mRNA 3’ UTR) could harbor multiple miRNA targets sites for many different miRNAs. As such, genes with extended 3’ UTRs are likely to be repressed by multiple miRNAs. This is likely the case for ABCA1, which has an unusually long (>3.3 kb) 3’ UTR, average is slightly over 1 kb, that makes it highly susceptible to miRNA targeting and post-transcriptional regulation. Not surprisingly, other miRNAs have been found to target ABCA1, including miR-758²⁶, miR-26²⁷, and miR-106²⁸. Using in silico prediction studies and profiling miRNAs that are altered in cholesterol-loaded peritoneal macrophages has proven to be successful strategy in identifying key miRNAs in cholesterol metabolism.²⁶. For example, miR-758 levels were found to be significantly decreased in cholesterol-loaded macrophages and were experimentally validated to directly target the ABCA1 3’ UTR²⁶. In a recent study, miR-26 was found to be suppressed by LXR activation and directly target the ABCA1 3’ UTR, and thus, repress cholesterol efflux and HDL-C levels²⁷. Strikingly, miR-106b was also found to directly target sites within the ABCA1 3’ UTR in neuronal cells, which was found to increase amyloid B secretion and production²⁸.

In this issue of Circulation Research, two independent research groups report that miR-144 directly targets the ABCA1 3’ UTR, thus repressing cholesterol efflux and HDL-C levels^{29, 30}. In one study, Carlos Fernandez-Hernando and colleagues (Ramirez et al.) present evidence that LXR activation up-regulates miR-144 transcription and that miR-144 directly targets ABCA1 and reduces cholesterol efflux in macrophages and the liver in part of a homeostatic network³⁰. In the other study, Peter Edwards and colleagues (de Aquiar Vallim et al.) found that FXR activation drives miR-144 transcription in the liver, which in turn directly targets ABCA1 and represses cholesterol efflux, thus promoting cholesterol excretion as bile acid²⁹. Together these studies support adding miR-144 to the list of miRNAs that regulate ABCA1 activity and thus cholesterol efflux from cells.

Although FXR-activation was found to only slightly increase miR-144 levels, previous studies have found that even small changes to ABCA1-targeting miRNAs (miR-33a/b) alter ABCA1 activity and cause large changes to HDL-C levels^{21-23, 31}. Most importantly, the promoter of miR-144/451 was found to harbor two functional FXR transcription factor binding sites. This work was aided by a recent FXR-chromatin immunoprecipitation (ChIP) sequencing study which identified a putative FXR binding site in the promoter of miR-144/-451³², and follow-up studies presented here identified a second FXR binding site in the proximal promoter, both of which were validated using promoter luciferase assays and site-directed mutagenesis. Using gene reporter (luciferase) assays, miR-144, but not miR-451, was found to directly target two sites within the mouse Abca1 3’ UTR. Most interestingly, tissue specific-FXR expression was used to demonstrate the requirement of hepatic FXR in miR-144 activation and FXR-mediated reduction in plasma cholesterol and HDL-C levels²⁹. Interestingly, FXR was found to up-regulate scavenger receptor BI (SR-BI) expression as well. Recently, over-expression of SR-BI was found to increase hepatic cholesterol and bile acid excretion when ABCA1 was inhibited³³. Thus, FXR-induced SR-BI levels may contribute to the observed decrease in HDL-C levels.

Although miR-144 levels are increased upon LXR activation in both macrophages and the liver, LXR is not predicted to directly bind to miR-144's promoter; rather this effect is likely mediated by a currently unknown indirect mechanism. The authors speculate that SREBP transcription factors, specifically SREBP1c, may play a role in the regulation. SREBP1c is a direct target of LXR and is predicted to bind to the promoter of miR-144/miR-451 at two sites. LXR-mediated SREBP transcription would also be predicted to promote miR-33a/b transcription. As such, miR-144 and miR-33a/b would work in tandem to repress ABCA1, and suggests that miR-33a/b targets ABCA1 in both low and high cholesterol conditions in the macrophage and liver. In ABCA1 gene reporter (luciferase) assays, miR-33a/b and miR-144 were each found to reduce luciferase activity individually; however, when combined they reduced luciferase activity by more than 60%. Nevertheless, their additive repression was not observed in cholesterol efflux assays when both were over-expressed in Huh7 cells, suggesting that while both are strong repressors of ABCA1, their biological effect is restricted to approximately 40% loss of ABCA1 function. These results may also indicate that ABCA1 expression and function is maintained by other mechanisms or that protein turnover accounts for this observation, and ABCA1 function could be more inhibited at later time points. Not surprisingly, inhibition of miR-33 or miR-144 did not affect the expression of the other. Further studies are required to resolve LXR-mediated transcriptional activation of miR-144 and the combinatorial effects of miR-33a/b and miR-144 in high sterol states. Nevertheless, the narrative of both articles is that sterol activation of miR-144 through nuclear receptor activation inhibits cholesterol efflux and HDL-C levels. The LXR-miR-144 network is distinct from the miR-33a/b-ABCA1 network in that miR-33 serves to prevent cellular sterol loss through efflux in low sterol states as the cell adapts to increase cellular cholesterol levels, whereas miR-144 prevents uncontrolled LXR-induced cholesterol efflux through ABCA1 in high sterol conditions.

Similar to miR-33a/b, miR-144 repression of ABCA1 is not cell type specific or restricted to macrophages, as this mechanism was found to exist in endothelial cells, macrophages, and hepatocytes. However, the effect of miR-144 on ABCA1 may be different in distinct tissues and cells²². For example, over-expression of miR-144 in the LXR agonist TO901317-stimulated human Huh7 cells only slightly reduced ABCA1 protein expression (approximately 12% loss), whereas in J774 mouse macrophages ABCA1 protein levels were substantially reduced (approximately 45% loss). Similar to many of the original miR-33a/b in vitro experiments^{21, 23}, the effects of miR-144 over-expression were observed in “TO901317-activated LXR” conditions. Nevertheless, de Aquiar Vallim et al. did observe an approximate 25% loss of ABCA1 protein and cholesterol efflux in non-stimulated mouse hepatocytes with miR-144 over-expression; and Ramirez et al. reported that inhibition of miR-144 increased ABCA1 protein levels in “non-stimulated” macrophages. In mice without LXR activation, over-expression and inhibition of miR-144 resulted in an approximate 20% change to HDL-C levels in both directions at day 6³⁰. These data suggest either that in some instances miR-144 repression of ABCA1 is more dramatic when ABCA1 is up-regulated or alternatively that miR-144 over-expression in enhanced by additional miR-144 transcription due to LXR agonists.

It is accepted that partial complementary seed-based binding of miRNAs to mRNAs results in both translational inhibition and mRNA degradation³⁴. However, a common observation between the miR-33a/b literature and these miR-144 studies is the curious lack of miRNA effect on ABCA1 mRNA abundance²¹. Over-expression of miR-144 was found to significantly reduce ABCA1 protein and cholesterol efflux in human hepatoma cells (Hep3B) and mouse primary hepatocytes, but failed to alter ABCA1 mRNA levels in both cell-types²⁹. In addition de Aquiar Vallim et al. only found ABCA1 protein to be affected with miR-144 over-expression in mice; however, Ramirez et al. did find miR-144 over-expression to modulate both ABCA1 mRNA and protein levels^{19, 29}. In addition, while activation of FXR with GSK2324 in mice did significantly reduce hepatic Abca1 mRNA levels, the effect was very modest and disproportionate to the effect on ABCA1 protein²⁹. Inhibition of miR-144 in basal non-stimulated Huh7 cells resulted in reduced hepatic miR-144 levels and elevated ABCA1 proteins levels, but also failed to affect (increase) ABCA1 mRNA levels³⁰. Interestingly, inhibition of miR-144 in mice was found to affect Abca1 mRNAs levels only in TO901317-treated mice; however, HDL-C levels were found to be significantly elevated in both TO901317-treated and untreated mice³⁰. The de Aquiar Vallim et al. study was consistent with this observation, as inhibition of miR-144 in mice only affected ABCA1 protein levels, which resulted in increased HDL-C levels. In addition to miR-33a/b and miR-144, ABCA1 also harbors a putative miR-27b target site within its 3’ UTR, and miR-27b over-expression in Huh7 cells also failed to reduce ABCA1 mRNA levels; however, neither protein or efflux assays were completed in this study¹⁷. Compensatory ABCA1 transcription or other indirect mechanisms may account for the lack of miR-144 or miR-33a/b effect on ABCA1 mRNA abundance, or these miRNAs could target genes (nucleases) responsible for mRNA degradation after translational inhibition. Nevertheless, it is clear that miR-144 and miR-33a/b directly target sites harbored within ABCA1's 3’ UTR.

To demonstrate the functional impact of miR-144 in vivo, both studies successfully used loss-of-function anti-miR approaches in mice. Recently, nucleic acid-based therapeutics have emerged as viable options to treat numerous diseases. siRNA drugs are designed to repress one specific gene with minimal off-target effects; however, miRNA–based strategies hold greater potential for manipulating larger cassettes of genes and possibly entire networks and pathways. Due to imperfect base-pairing associated with seed-based miRNA recognition and targeting of mRNAs, one miRNA has the potential to regulate complete biological pathways through repressing multiple genes with in the pathway. Recently, the diversity of miRNA inhibitors has rapidly expanded to include many different RNA chemistries aimed at increasing stability, efficacy, target affinity, and tissue localization. These include single-stranded locked-nucleic acids (LNAs), 2’-methoxymethyl (2’MOE), 2’-fluoro-methoxymethyl (2’F/MO), and 2’O-ethyl (cEt) modifications. The most widely studied and tested anti-miR is Miravirasen (Santaris Pharma), an anti-miR-122 drug that has advanced to phase 2 clinical trials for the treatment of hepatitis C virus. Many other anti-miRNA-targeting strategies have been tested in preclinical studies, including anti-miR-103/miR-107 (insulin sensitivity), anti-miR-208 (cardiac dysfunction), and anti-miR-29 (fibrosis). Most relevant, anti-miR-33 approaches have been shown to increase ABCA1 protein and raise HDL-C levels in mice and non-human primates^{19, 20} and promote reverse cholesterol transport and reduce atherosclerosis in mice²⁰.

In both studies presented here, inhibition of miR-144 was found to increase ABCA1 expression and function, and raise HDL-C levels in mice. Each study used a slightly different approach to inhibit miR-144 in vivo. de Aquiar Vallim et al. used miR-144 2’F/MOs (Regulus Therapeutics) to inhibit endogenous miR-144 levels in wild-type c57Bl/6 mice on a high-fat diet though bi-weekly treatments (intraperitoneal injections) of 5 mg/kg anti-miR-144 for 4 weeks²⁹. Ramirez et al. used mimetics and inhibitors (mirVana, 7 mg/kg) coupled with Invivofectamine (Invitrogen) for intravenous injections twice every 3 days³⁰. In previous studies, anti-miR-33 (2'F/MO) was injected (10 mg/kg) subcutaneously to demonstrate the ability of anti-miR-33 therapy to raise HDL-C levels and regress atherosclerosis^19-21. Based on results from the studies presented here, anti-miR-144 therapy might also be explored as a nucleic acid-based therapy to increase HDL-C levels and, potentially, to reduce atherosclerosis.

Small RNAs, namely miRNAs, are found in all extracellular compartments and biological fluids³⁵. Extracellular miRNAs are remarkably stable, likely due to their association with lipid and protein complexes, including HDL³⁶. Most interestingly, miR-451 resides on the same genomic (poly-cistronic) cluster and is co-transcribed with miR-144; and miR-451 is one of the most highly exported miRNA from cells. Although miR-144 and miR-451 are co-transcribed, miR-144 is rarely exported from cells. Using real-time PCR based methods, miR-33a/b, miR-144, and miR-758 were not detected on HDL³⁶; however, other ABCA1-regulating miRNAs (miR-26 and miR-106b) and miR-451 are consistently found on HDL in humans and mice. Nevertheless, using high-throughput small RNA sequencing to detect and quantify extracellular miRNAs, miR-33, miR-144, and miR-451 were found on HDL in specific samples (unpublished data).

In summary, these studies highlight a common theme with metabolic miRNAs—that miRNAs serve as biological rheostats and buffers in metabolism through feedback (LXR) and directional flux (FXR) networks. The key message supported by these studies is that nuclear receptors control cholesterol efflux through miR-144 and post-transcriptional regulation of ABCA1. LXR and FXR-mediated repression of ABCA1 through post-transcriptional regulation via miR-144 is a fascinating glimpse of the complexity in cellular cholesterol metabolism. Furthermore, these studies highlight the potential for modulating directional cholesterol flux through nucleic acid therapy-based inhibition of miR-144.