Biogenesis and mode of action
miRNAs are initially transcribed as long precursor molecules termed primary miRNAs (Fig. 3). These are synthesized either by RNA polymerase II transcription or from the cleavage of introns in protein-coding genes . Primary miRNAs are then processed in a two-step sequence via the actions of Drosha and Dicer, Rnase III enzymes. First, nuclear Drosha processes the primary miRNA into an approximately 70-nucleotide precursor hairpin (precursor miRNA), which undergoes cytoplasm export via Exportin 5. In the cytoplasm, the precursor miRNA is further processed by Dicer into an approximately 21-nucleotide miRNA/miRNA* duplex, which will form a complex with the argonaut (AGO) proteins to form the miRNA-induced silencing complex (miRISC). One miRNA strand acts as the active strand, whereas the other one, also known as passenger strand or miRNA*, is released and, in most cases, degraded.
Figure 3. miRNA synthesis and mechanism of action. miRNAs are processed from precursor molecules (pri-miRNA). The pri-miRNAs can be produced by two different mechanisms, including the action of RNA polymerase II transcription or the cleavage of introns from protein-coding genes. Next, pri-miRNAs are processed by Drosha in the nucleus into an approximately 70-nucleotide precursor hairpin (pre-miRNA), which is exported to the cytoplasm via Exportin 5. In the cytoplasm, pre-miRNA is processed by Dicer into an approximately 21-nucleotide miRNA/miRNA* duplex from where the mature miRNA will then form a complex with AGO and GW182 (miRISC). The other strand (passenger strand or miRNA*) is released and often degraded. The majority of the miRNAs base-pair imperfectly with sequences in the 3′-UTR of target mRNAs, which inhibits protein synthesis by either repressing translation or promoting mRNA deadenylation and decay. Deadenylation of mRNAs is mediated by GW182 proteins, which interact with AGOs and act downstream. When miRISC-containing AGO2 encounters mRNAs bearing sites almost perfectly complementary to miRNA, these mRNAs are cleaved endonucleolytically and degraded.
Download figure to PowerPoint
Efficient mRNA targeting by miRNAs requires base-pairing with the seed region of the miRNA, a region formed by two to eight nucleotides at the 5′ end of an miRNA . Most miRNAs bind with an imperfect base-pair match with the sequences in the 3′-UTR of target mRNAs, thereby inhibiting protein synthesis by repressing translation or inducing mRNA deadenylation. Although the mechanisms by which miRNAs induce translational repression are diverse and still not fully understood, miRNA-mediated mRNA deadenylation is more or less established, being mediated by GW182 proteins, upon interaction with AGOs. Although the N-terminal region of GW182 interacts with AGO via its GW repeats, the C-terminal region interacts with the poly(A) binding protein and recruits deadenylases . miRNA-induced mRNA deadenylation may ultimately lead to the decay of target mRNAs via the recruitment of decapping machinery. Finally, when AGO2 is present in the miRISC and it encounters mRNAs that have sites almost matching to miRNA, these mRNAs are degraded by endonuclear cleavage, a typical mechanism in plants .
miRNAs as crucial regulators of NAFLD pathogenesis and progression
Because of their crucial role in lipid metabolism, cell growth, differentiation, apoptosis and inflammation, miRNAs are now being extensively studied as important regulators of NAFLD pathogenesis (Table 1). Several microarray or profiling experiments have already shown that miRNAs are differentially expressed in human NASH, as well as after genetic- and/or diet-induced mouse models of NASH . Furthermore, NASH severity and susceptibility to NASH could be determined by variations in miRNA expression response . For example, in ob/ob mice, 11 miRNAs were found to be deregulated in NAFLD compared to control mice. Among these, eight miRNAs (miR-34a, -31, -103, -107, -194, -335-5p, -221 and -200a) were upregulated and three miRNAs (miR-29c, -451 and -21) were downregulated . NAFLD pathogenesis in the low-density lipoprotein receptor knockout mice is also associated with modulation of the hepatocyte miRNA profile, including a five-fold decrease in miR-216 and miR-302a expression . Another study, using different diet-induced NASH models in mice, has also identified between 1% and 3% changes in miRNA expression profiles during steatohepatitis, including variations in miR-705, -1224, -182, -183 and miR-199a-3p . In addition, rats fed either a standard diet with high fructose, a HFD or a diet with high levels of fats and fructose, all showed significant downregulation of miR-122, -451 and miR-27, as well as upregulation of miR-200a, -200b and miR-429 . One of the first reports in human NASH livers identified 46 miRNAs to be under- or overexpressed compared to control samples . Taken together, these findings support the participation of miRNAs in the pathophysiological processes of NAFLD.
Table 1. Summary of dysregulated miRNAs in NAFLD.
|Humans||miR-26b; miR-28; miR-30d; miR-92b; miR-122;miR-126; miR-139; miR-145; miR-188; miR-191*; miR-198; miR-203; miR-223; miR-296-5p; miR-361; miR-375; miR-563; miR-574; miR-601; miR-617; miR-641; miR-671; miR-765; miR-768-5p||miR-15b[155b]; miR-16[39,152b]; miR-21[39,154b]; miR-23a; miR-23b; miR-24; miR-27b; miR-34a[2,39,152b,154b]; miR-99b; miR-100; miR-122[152b,154b]; miR-125b; miR-127; miR-128a; miR-128b; miR-146b; miR-181b; miR-199a; miR-199a*; miR-200a; miR-214; miR-221[39,129]; miR-222[39,129]; miR-224; miR-451[154b]; miR-455|
|Animal models||miR-15b; miR-21; miR-27; miR-29c[111,112]; miR-33a[123a]; miR-122[18a,111,115,117a,130,151]; miR-155[131a]; miR-192 miR-203; miR-216; miR-302a; miR-467b; miR-451[112,115]||miR-21; miR-31; miR-33a[120a,122]; miR-34a[111,112,134,137,137a,140,140a,151,151b]; miR-103; miR-107; miR-122[147b,150b,151b]; miR-155[111,130]; miR-181a[151,151b]; miR-182; miR-183; miR-192[151b]; miR-194; miR-199a-3p; miR-200a[112,115]; miR-200b[111,115,151]; miR-221[111,112,130,151,151b]; miR-222; miR-335-5p; miR-429; miR-705; miR-1224|
miR-122 is a liver specific miRNA playing a key regulatory role in lipid and fatty acid metabolism, as well as cholesterol accumulation. miR-122 is a post-transcriptional regulator of CYP7A1, the rate-limiting enzyme controlling bile acid synthesis in human hepatocytes . Recent studies showed that miR-122 knockout mice accumulate triacylglycerol in the liver, as a result of the upregulation of enzymes responsible for triacylglycerol synthesis and storage. In addition, miR-122 knockout mice also display hepatic inflammation, progressive fibrosis and, ultimately, hepatocellular carcinoma . miR-122 also regulates fibrogenic factors, including Kruppel-like factor 6 that targets transforming growth factor β1. As such, when inhibited, miR-122 leads to activation of hepatic stellate cells and fibrogenic processes. miR-122 knockout mice show evidences of steatosis and abnormal levels of very-low-density lipoproteins and high-density lipoproteins .
miR-33a plays a key role in bile acid synthesis, fatty acid oxidation and cholesterol homeostasis [118, 119]. In particular, when cellular cholesterol levels decrease, miR-33a expression is co-induced with sterol regulatory element-binding protein 2 (SREBP2) mRNA. miR-33a silencing promoted regression of atherosclerosis in mice, which suggests that miR-33a acts in synergy with SREBP2 to regulate cholesterol homeostasis . A recent study showed that cholesterol might repress miR-33a levels to increase CYP7A1 expression, as well as cholesterol efflux transporters. By contrast, SREBP2 and miR-33a activation downregulate cholesterol efflux transporters and bile acid synthesis, which results in increased intrahepatic cholesterol ; as such, any imbalance in this regulatory circuit will increase hepatocyte lipid content and, ultimately, induce NAFLD. This appears to be the case in hepatic stellate cells exposed to free cholesterol; suppression of PPAR γ signalling accompanying hepatic stellate cells activation was shown to enhance both SREBP2 and miR-33a signalling. As a consequence, there is greater accumulation of free cholesterol in hepatic stellate cells, further sensitizing them to TGFβ-induced activation in a vicious cycle, leading to exaggerated liver fibrosis in NASH . SREBP1 is also a target of miR-33, and is significantly enhanced in miR-33 knockout mice. Unexpectedly, these mice develop obesity and liver steatosis by a SREBP1-dependent mechanism. It was suggested that the sustained, continuous absence of miR-33 in the liver may have detrimental effects as opposed to its temporal inhibition in pathological settings . miR-33a and -33b have been also described to inhibit genes involved in fatty acid metabolism and insulin signalling in hepatocytes , further implying their pathogenic role during NAFLD.
miR-296-5p was recently identified as a negative regulator of PUMA expression during hepatocyte lipoapoptosis. In NASH patients, hepatic miR-296-5p levels are reduced and associated with increased PUMA expression, confirming an inverse association between both . These results suggest that lipoapoptosis, a key mediator of liver injury in NAFLD, results at least in part from miR-296-5p downregulation. In turn, lipogenesis is targeted by miR-1 and miR-206. These miRNAs directly repress LXRα, thereby inhibiting the expression of LXRα target genes, such as SREBP1c, fatty acid synthase, carbohydrate responsive element-binding protein and acetyl-CoA carboxylase 1. Accodingly, miR-1/miR-206 attenuates LXRα-induced lipogenesis, hepatic steatosis and NASH .
miR-21 is intrinsically connected with NAFLD because it represents a crucial factor affecting PTEN expression during the metabolic syndrome, both in the human liver and in primary human hepatocytes. Indeed, excessive circulating unsaturated FFAs, such as oleic and palmitic acids, increase miR-21 expression, resulting in PTEN downregulation and the development of steatosis . Liver-specific PTEN knockout mice develop hepatic steatosis, inflammation and fibrosis, all constituting biochemical and histological evidence of NASH . Regarding fibrosis, it was also recently described that miR-221 and -222 are upregulated in the human NAFLD liver, in a fibrosis progression-dependent manner . PTEN is also a direct target of miR-155; miR-155, as well as miR-221/222 and miR-21, were found to be upregulated at early stages of NASH-induced hepatocarcinogenesis, whereas miR-122 was downregulated. Expression of miR-155 correlated with PTEN protein levels and with diet-induced histopathological changes in the liver . miR-155 knockout mice fed a HFD develop hepatic steatosis in parallel with increased hepatic expression of genes involved in glucose regulation, fatty acid uptake and lipid metabolism. In addition to miR-1 and miR-206, miR-155 also directly targets LXRα . It may be that increased hepatic expression of miR-155 is relevant in triggering hepatocarcinogenesis, rather than strongly contributing to NAFLD progression.
IR in NAFLD also appears to correlate with miRNA changes; miR-467b was shown to be downregulated in liver tissues of HFD-fed mice and in steatosis-induced hepatocytes, resulting in upregulation of hepatic lipoprotein lipase, a direct target of miR-467b. Moreover, the miR-467b/LPL axis is associated with IR . Furthermore, PPARα mediates IR and lipolysis by FFAs in the liver. In that regard, miRNA-10b has been shown to regulate cellular steatosis level by targeting PPARα .
The exacerbation of inflammation observed at more severe stages of NAFLD increases p53 expression, which mediates mitochondrial pathways of apoptosis. Because p53 is a key modulator of steatosis, its regulatory control appears to fail in NAFLD because activated p53 upregulates pro-apoptotic miR-34a . Moreover, activation and stabilization of p53 results in a feed-forward regulatory mechanism to activate downstream genes involved in apoptosis, oxidative stress and IR . In agreement, pharmacological inhibition of p53 attenuates hepatic steatosis and liver injury in mice fed a HFD . miR-34a is the most upregulated miRNA in the livers of mice fed with a HFD, as well as in patients with metabolic syndrome and NASH [2, 39]. A recent study has shown that aberrantly elevated miR-34a in obesity attenuates hepatic FGF19 signalling by directly targeting co-receptor β-Klotho, thereby severely impairing postprandial responses and contributing to the metabolic syndrome . Moreover, the main target of miR-34a, sirtuin 1 (SIRT1), has a central role in regulating hepatic fatty acid metabolism; it abolishes ectopic fat accumulation by inducing fatty acid β-oxidation and decreases de novo fatty acid synthesis. Accordingly, hepatocyte-specific SIRT1 knockout mice fed with a HFD display significant levels of hepatic steatosis and inflammation . In addition, morbid obese patients with NAFLD showed increased miR-34a expression, p53 acetylation and apoptosis, as well as decreased SIRT1 expression in more severe stages, suggesting that this pathway contributes to disease progression . Furthermore, SIRT1 silencing in adipocytes inhibits insulin-stimulated glucose uptake and glucose transporter type 4 translocation, tyrosine phosphorylation of IRS-1, and phosphorylation of AKT and extracellular signal-regulated kinases, accompanied by increased phosphorylation of JNK and serine phosphorylation of IRS-1. By contrast, SIRT1 activation increases glucose uptake and insulin signalling and decreases serine phosphorylation of IRS-1 . SIRT1 activity is additionally targeted by hepatic miR-34a via nicotinamide phosphoribosyltransferase (NAMPT), the rate-limiting enzyme for NAD+ biosynthesis . In particular, miR-34a overexpression in hepatocytes reduces NAMPT/NAD+ levels, increases acetylation of SIRT1 targets PPAR γ coactivator 1-α, SREBP-1c, farnesoid X receptor and NF-κB, and results in steatosis, inflammation and glucose intolerance. As such, in parallel with the miR-34a/SIRT1/p53 axis, the miR-34a/NAMPT axis presents a potential target for treating NAFLD. Finally, miR-34a inhibition of SIRT1 has been associated with downstream dephosphorylation of AMP kinase and hydroxy-3-methyl-glutaryl-CoA reductase , suggesting that miR-34a-dependent deregulation of cholesterol metabolism may also contribute to NAFLD progression and severity.
The miRNA machinery is also affected in NAFLD. In a recent study, analyzing the visceral adipose tissue from 24 morbidly obese patients, it was shown that the miRNAs processing enzymes Dicer and Drosha increase with NAFLD severity and, furthermore, Dicer expression also correlates with hepatocyte ballooning degeneration .
miRNAs as biomarkers in NAFLD
In addition to the significant differences between miRNA expression profiles, when comparing the diseased liver versus normal states, miRNAs are also found to be deregulated in the body fluids of patients with different liver diseases. Indeed, miRNAs can be detected in several fluids, such as tears, seminal plasma, cerebrospinal fluid and pleural fluid. For example, in urine, saliva and plasma samples, miRNA expression and concentration has already been associated with the presence of different human tumours . Circulatory miRNAs are not usually found in an isolated form in plasma but, instead, are coupled with the AGO2 ribonucleoprotein complex  or inside microvesicles . In addition, plasma miRNAs may also be transported in association with high-density lipoproteins, allowing their delivery to recipient cells . This may constitute a new mechanism of miRNA communication between different tissues and a new set for miRNA-targeting. For these reasons, and because miRNA expression profiles are typically tissue- and disease-specific, miRNAs are being increasingly studied as diagnostic and biomarker tools. Because the only way to detect NAFLD is by liver biopsy, effective serum biomarkers have been eagerly warranted. It is now apparent that plasma miRNA profiles may reflect and differentiate between several types of liver injury, namely from acute to chronic liver injury or from steatosis to NASH, fibrosis and, ultimately, hepatocellular carcinoma . Overall, the most studied liver miRNA as a possible biomarker and means of diagnosis is miR-122. For clinical relevance, studies have compared miR-122 serum levels analysis with ALT activity, a currently used marker of liver damage. Under physiological conditions, ALT activity is approximately 3000-fold higher in the liver than in serum, whereas miR-122 concentrations are several thousand-fold higher in the liver compared to other tissues. Therefore, any increase in miR-122 serum concentrations will be higher than that observed for ALT . In addition, miR-122 shows reasonable sensitivity and specificity in animal models and is easily detectable in blood samples . Altogether, miR-122 serum levels account as a sensitive biomarker for the early detection of hepatotoxicity. Moreover, miR-122 levels correlate with histopathological changes in the liver more sharply than ALT activity; in acute liver injury models, for example, miR-122 plasma levels increase much faster and dramatically than plasma ALTs, reflecting the extent of hepatocellular injury . In particular, miRNAs have been shown to correlate with the extent of NAFLD-associated liver injury in mice . In different NAFLD mice models, levels of circulating miR-34a, -122, -181a, -192 and -221 significantly correlated with the severity of NAFLD-specific liver pathomorphological features , suggesting their use as non-invasive biomarkers for monitoring the extent of NAFLD-associated liver injury, as well as susceptibility to NAFLD. Several studies using human samples appear to support this idea; significantly higher serum levels of miR-122, -34a and miR-16 were found in NAFLD patients compared to controls, with miR-122 and miR-34a levels positively correlating with disease progression from simple steatosis to NASH, fibrosis stage and inflammation, as well as with liver enzymes. miR-122 further correlated with serum lipids levels . Serum levels of miR-122, -572, -575, -638 and -744 were also found to be deregulated in patients with NASH or chronic hepatitis B . Finally, a very recent profiling study demonstrated that patients with NAFLD have higher serum levels of miR-21, -34a, -122 and -451, where the serum level of miR-122 further correlated with the severity of liver steatosis . miR-15b, which was already suggested to play a pathogenic role in NAFLD, is also elevated in the serum of NAFLD patients compared to healthy subjects . However, a word of caution is still necessary because circulating miRNAs levels in humans may vary according to age, sex and race, and may also be influenced by extrinsic factors, such as temperature .
miRNA-based therapy for liver diseases
Deregulated miRNAs in liver diseases can and should be considered as potential therapeutic targets. When the aim is to inhibit one particular miRNA expression, several hypotheses have been raised and include the use of miRNA antagonists such as antisense oligonucleotides locked-nucleic acids or antagomirs, ‘miRNA sponges’, ‘miRNA decoys’ and ‘miRNA erasers’. Chemically modified antisense oligonucleotides were the first to be described and reported as inhibiting miRNAs in cultured cells and invertebrates . However, they are no longer studied intensively. In turn, antagomirs were the first miRNA inhibitors to achieve strong, positive results in mammals and comprise a cholesterol-conjugated RNA of approximately 419 nucleotides in length, for highest efficiency . These antagonists were primarily used and studied in targeting miR-122 in the liver, thus regulating the cholesterol synthesis pathway .
As for most short-interference RNA techniques, there are still some barriers to overcome before the use of miRNAs can be established therapeutically. miRNA molecules are easily destroyed by endogenous RNAses. In addition, miRNAs affect several pathways in various organs, which require highly developed systems of drug delivery to avoid adverse effects. To overcome these barriers, several solutions have been proposed. Namely, several chemical changes have been developed to improve the characteristics of antagomirs. Locked nucleic acid modified-antimiRNAs, for example, have the ability to inhibit miRNAs in vivo and offer several advantages over conventional antagomirs, such as a low amount of off-target effects . ‘miRNA sponges’ consist in a set of competitive inhibitors that inhibit miRNAs with a complementary heptameric seed analagous to a sponge that blocks miRNA activity. This mechanism showed equal efficiency with respect to depressing miRNA expression compared to miRNA antagonists . On the other hand, ‘miRNA erasers’ work via a mechanism similar to ‘miRNA sponges’ but with less stem-loop sequences and their delivery has to be made via a viral vector . Finally, ‘miRNA decoys’ consist of a set of sequences that will bind to the targets of a specific miRNA, thus preventing it from attaching and thus repressing that pathway. The first miRNA decoy consisted of an adenoviral vector that bound to two targets of miR-133 . Miravirsen is a good example of an antisense miRNA-based therapy in the liver, already undergoing clinical trials, namely for hepatitis C. Miravirsen is a locked nucleic acid-modified oligonucleotide (anti-miR-122) capable of suppressing HCV viraemia with no evidence of side effects or viral resistance. It is easily delivered into the liver by intravenous injection [164, 165]. The likelihood of treatments using miRNAs in liver disease represents a great advantage over other organ diseases because these oligos are first distributed preferentially to the liver . Furthermore, this field is expanding significantly, as highlighted by a recent report demonstrating that lipid-based nanoparticles containing oleic acid significantly enhance the delivery efficacy of miR-122 into the liver .
When the objective is to increase the expression of a specific miRNA, miRNA mimics are used. For example, the systemic administration of miR-26a using an adenoassociated virus was the first miRNA-replacement strategy for hepatocellular carcinoma. miR-26a delivery resulted in tumour-specific apoptosis by direct targeting of cyclins D2 and E2 . More recently, miR-34a has been delivered to NOV340 hepatocellular carcinoma cells via liposomes. The animal group treated with NOV340/miR-34a showed no side effects and demonstrated prolonged survival compared to untreated animals .
Concerning NAFLD, very few studies have yet uncovered the full potential of miRNAs as effective therapeutic agents. An interesting, proof-of-concept study has been performed in African green monkeys, where the systemic delivery of an anti-miRNA oligonucleotide targeting both miR-33a and miR-33b was shown to increase the hepatic expression of miR-33 target genes involved in fatty acid oxidation and to reduce the expression of genes involved in fatty acid synthesis. As a result, plasma high-density lipoprotein levels were increased, whereas very-low-density lipoprotein-associated triglycerides were decreased . In a model that is extremely relevant for humans, these findings provide an indication that the successful therapeutic targeting of select miRNAs might prove highly beneficial for treating the metabolic syndrome and associated pathologies, such as NAFLD.