Non-coding RNA Crosstalk with Nuclear Receptors in Liver Disease

Abstract

The dysregulation of nuclear receptors (NRs) underlies the pathogenesis of a variety of liver disorders. Non-coding RNAs (ncRNAs) are defined as RNA molecules transcribed from DNA but not translated into proteins. MicroRNAs (miRNAs) and long non-coding RNAs (lncRNAs) are two types of ncRNAs that have been extensively studied for regulating gene expression during diverse cellular processes. NRs as therapeutic targets in liver disease have been exemplified by the successful application of their pharmacological ligands in clinics. MiRNA-based reagents or drugs are emerging as flagship products in clinical trials. Advancing our understanding of the crosstalk between NRs and ncRNAs is critical to the development of diagnostic and therapeutic strategies. This review summarizes recent findings on the reciprocal regulation between NRs and ncRNAs (mainly on miRNAs and lncRNAs) and their implication in liver pathophysiology, which might be informative to the translational medicine of targeting NRs and ncRNAs in liver disease.

1. Introduction

Nuclear receptors (NRs) constitute the largest superfamily of transcription factors (TFs) and play pivotal roles in many biological processes, including metabolism, reproduction, development, circadian rhythms, immunity, and senescence. In response to ligand binding, a NR will allosterically change its conformation, followed by oligomerization, binding to responsive DNA elements, and dissociation of corepressors and/or association of coactivators, ultimately resulting in the alteration of transcription to exert its biological function [1, 2]. Ligands for most NRs (“adopted NRs”) have been identified, such as hormones, fatty acids, bile acids, bilirubin, xenobiotics, and metabolic intermediary molecules. However, a number of NRs, referred to as orphan receptors, have no known physiological or endogenous ligands [2]. So far, 48 NRs encoded in the human genome and 49 in the mouse genome have been found [1, 3]. Due to the alternative splicing, the number of functional NRs is larger than the literary number. For instance, constitutive androstane receptor (CAR, NR1I3) has 35 splice variants, and at least 16 isoforms of consensus coding sequence have been recognized in humans. The isoforms have been suggested to serve as dominant-negative inhibitors of NRs [4]. Despite the diversity in size, shape, and ligand selectivity, almost all NRs share a common arrangement of molecular domains (Figure 1). Detailed perspectives on the structure and domain function of NRs have been reviewed elsewhere [5, 6].

Figure 1.

The molecular domains of typical NRs. The overall architecture for a typical NR is composed of five domains: (A/B) an N-terminal regulatory domain (NTD) that displays a large disparity between NRs, is targeted by multiple post-translational modifications (PTMs), and contains a powerful transactivation subdomain, activation function 1 (AF-1); (C) a DNA binding domain (DBD) that is the most conserved in comparison with other domains; (D) a hinge region that has the least conservation and can host PTM sites and nuclear localization signals; (E) a ligand-binding domain (LBD) that allows NRs to recognize and bind respective ligands, interact with coregulator proteins, and partner to form homo- or heterodimers; the C-terminal part of the LBD is shown to have a ligand-inducible activation function, termed AF-2; (F) a highly variable C-terminal domain.

NRs maintain the homeostasis of normal liver function; dysregulation and genetic variance of NRs contribute to the development and progression of multiple liver pathologies, such as steatosis, fibrosis, cholestasis, drug-induced liver injury, and neoplasia [7]. The concept of targeting NRs to develop effective therapies has succeeded in drug development. A dual peroxisome proliferator-activated receptor (PPAR) agonist, saroglitazar, binding to both the α and γ PPAR isoforms, has been approved for the treatment of metabolic syndrome in India by the Drug Controller General of India and undergone a phase II clinical trial for the treatment of non-alcoholic fatty liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH) in the US [8]. Obeticholic acid (OCALIVA®), the first farnesoid X receptor (FXR) agonist, approved for the treatment of primary biliary cholangitis (PBC), is undergoing development for the treatment of several liver diseases and related disorders [9, 10]. Despite the advancement in our understanding of NRs, the regulation and function of NRs, especially those of orphan NRs, are still elusive in the pathogenesis of liver disease. The regulatory networks of NRs are complex, which include ligands, co-regulators (coactivators/corepressors), protein post-translational modifications (PTMs), and non-coding RNAs (ncRNAs). NcRNAs regulate NRs to modulate a broad range of biological functions, including metabolism, inflammation, proliferation, apoptosis, differentiation, tissue remodeling, and organ crosstalk, contributing to the development of varieties of liver disease [11, 12].

The transcripts for ncRNAs account for more than 80% of transcriptomes in mammalian cells [13, 14]. According to their length, ncRNAs can be arbitrarily grouped into short ncRNAs (sncRNAs, < 200 bp) and long ncRNAs (lncRNAs, > 200 bp). SncRNAs include the highly abundant transfer RNAs (tRNAs), ribosomal RNAs (rRNAs), small nuclear RNAs (snRNAs), and small nucleolar RNAs (snoRNAs), as well as small regulatory RNAs such as microRNAs (miRNAs/miRs), short interfering RNAs (siRNAs), and Piwi-interacting RNAs (piRNAs) [15, 16]. LncRNAs are largely tissue-specifically expressed in comparison with miRNAs and participate in gene transcriptional, post-transcriptional, and epigenetic regulation [17]. The latest release of miRBase (v22.1) (October 2018) (http://www.mirbase.org) reports 38,589 entries of hairpin precursor miRNAs which give rise to 48,860 mature miRNA products; 1,917 precursors and 2,654 mature miRNAs in humans have been annotated in this repository [18]. The human lncRNA database LNCipedia (https://lncipedia.org) reports that the lncRNA gene loci in the human genome are around 55,000, and 1,555 lncRNA genes are currently annotated with functional information [19]. Several hub miRNAs in biological processes such as lipid and glucose metabolism have been described [20]. In a comprehensive review, up to 60 lncRNA genes related to lipid metabolism and the different action mechanisms of lncRNAs have been elaborated [21]. The full annotation of ncRNAs on their roles in biological processes and human diseases is still incomplete.

In this review, we summarized the mutual regulation between ncRNAs (mainly miRNAs and IncRNAs) and NRs on the pathogenesis of liver diseases, particularly NAFLD, liver fibrosis, liver cancer, cholestatic liver disease, and drug-induced liver injury (Table 1).

Table 1.

NRs are regulated by ncRNAs in liver pathophysiology.

NRs	Regulatory ncRNAs	Target validation	Models (cells/tissues)	Pathophysiological processes or liver diseases
TRβ	miR-383, −146b [45, 46]	−146b	Liver tissues and primary hepatocytes from high-fat diet-fed mice	Steatosis and lipid metabolism in NAFLD
	miR-27a, -181a, -204 [49]	Yes	Rat HCC tissues and hepatic cell lines	HCC progression
PPARα	miR-21 [61, 62]	Yes	Liver tissues and primary cells	Liver injury, inflammation, and fibrosis in NASH
	miR-17-5p [63]	Yes	MiR-17 transgenic mouse liver tissues	Hepatic steatosis in NAFLD
	miR-27b [64]	Yes	Cell lines and human liver tissues	N.D.
	miR-34a [65]	Yes	Mouse liver tissues	Lipid metabolism in NAFLD
	miR-540 [66]	Yes	Gfer−/− mouse liver tissues	Steatosis and fibrosis in NAFLD
	miR-200 family and miR-34a [67]	Yes	Ire1α−/− and high-fat diet-fed mice, primary hepatocytes, and human liver tissues	Hepatic steatosis in NAFLD
	miR-10b [68]	Yes	Hepatic cell lines	Lipid metabolism and steatosis
	miR-155 [70, 247]	Yes	Chronic ethanol feeding and CCI4 treatment in mice	Steatosis and inflammation in ALD and liver fibrosis
	miR-9 [71, 72]	Yes	HCC tissues and hepatic cell lines	Cell invasion, proliferation, and lipid metabolism in HCC
	miR-27a [248]	Yes	Human hepatoma cells	Lipid metabolism and HCV replication
	circRNA_0046366 [73]	N.A.	Hepatic cell lines	Steatosis
	HULC [72]	N.A.	HCC tissues and hepatic cell lines	Lipid metabolism in HCC
PPARβ/δ	miR-9 [80]	Yes	Gestational diabetes mellitus (GDM) rat livers	Lipid metabolism in fetal livers of GDM pregnancies
	miR-122 [106]	Yes	Mouse livers	Circadian control of lipid metabolism in the liver
PPARγ	miR-27b [77]	Yes	HSCs (LX-2 cells)	HSC activation
	miR-27b [78]	Yes	Macrophages	Inflammatory response
	miR-27b [79]	Yes	Zebrafish	Hyperlipidemia and hepatic adipogenesis
	miR-130 [80]	Yes	Gestational diabetes mellitus (GDM) rats	Lipid metabolism in fetal livers
	miR-34a, -34c [82]	Yes	Rat primary HSCs and HSC cell lines	HSC activation and live fibrosis
	miR-130a, -130b [83]	Yes	Rat HSC cell lines (HSC-T6)	HSC activation and liver fibrosis
	miR-942 [84]	Yes	Patients liver tissues and human primary HSCs and HSC cell lines	HSC activation in HBV liver fibrosis
	miR-155 [70]	Yes	Chronic ethanol feeding and CCI4 treatment in mice	Steatosis and inflammation in ALD and fibrosis
	miR-29a [85]	No	Bile duct-ligation (BDL) in mice	HSC activation and cholestatic liver fibrosis
	miR-128-3p [86]	Yes	Primary mouse HSCs and HSC cell lines	HSC activation and live fibrosis
	miR-130b [89]	Yes	HCC tissues and hepatic cell lines	EMT, cancer cell migration and invasion in HCC
	miR-27a [90, 91]	Yes	HCC tissues, hepatic cell lines, and HepG2 xenografts	Proliferation in HCC
REV-ERBα	miR-140-5p, -185-5p, -326-5p, -328-5p [102]	Yes	Mouse livers	Hepatic steatosis
RORα	miR-1246 [104]	Yes	HCC cells and tumors	HCC progression
LXRα	miR-27a/b-3p [110]	N.D.	Cultured HSCs and mouse liver tissues	HSC activation and liver fibrosis
	miR-613 [112]	Yes	Human hepatic cells	N.D.
	miR-613 [113]	Yes	THP-1 macrophages	Cholesterol metabolism
	miR-613 [114]	Yes	HepG2 cells	Lipogenesis
	miR-155 [115]	Yes	Mouse and human livers and hepatic cells	Lipid metabolism in NAFLD
	miR-1, -206 [116]	Yes	Hepatocytes	Lipogenesis
	miR-206 [117]	Yes	Hepatic cells and macrophages	Cholesterol efflux in macrophages
LXRβ	miR-7 [94]	Yes	Hepatic cell lines	Lipid metabolism
FXR	miR-421 [129, 130]	Yes	Hepatic cancer cell lines	Proliferation, migration, and invasion in BTC and HCC
	miR192 [131]	Yes	Hepatic cancer cell lines	Cell proliferation in HCC
	miR-194 [126]	Yes	Mouse liver tissues and cultured hepatocytes	Lipid metabolism in NAFLD
	miR-122 [126, 127]	Yes	Mouse liver tissues	Hepatic triglycerides levels in NASH
	IncLSTR [11]	N.A.	Mouse livers, plasma, and primary hepatocytes	Systemic lipid homeostasis
VDR	miR-125a-5p [142]	Yes	HCC tissues and cells	Malignancy of HCC
	miR-351 [143]	Yes	HSCs and mouse liver tissues	Liver fibrosis
PXR	miR-148a [146]	Yes	Hepatic cell lines and human liver tissues	CYP3A4 expression
	miR-148a [147]	Yes	Hepatic cell lines and patient serum	Estrogen-induced cholestasis
	miR-18a-5p [144]	Yes	Primary human hepatocytes and hepatic cell lines	CYP3A4 expression
	LINC00844 [148]	N.A.	Human liver tissues and HepaRG cells	Drug metabolizing enzymes expression in response to APAP toxicity
	HNF1α-AS1	N.A.	Hepatic cell lines and human liver tissues	P450 expression
CAR	miR-137 [161]	N.D.	Mouse livers and primary hepatocytes	Bilirubin clearance
	miR-122 [156, 157]	N.D.	Hepatic cell lines	Induction of drug-metabolizing enzymes
HNF4α	LINC00844 [148]	N.A.	Human livers and HepaRG cells	Drug metabolizing enzymes expression in response to APAP toxicity
	miR-24, miR-629 [168]	Yes	HCC cell lines and tumors	Liver tumorigenesis and inflammation
	miR-21 [169]	Yes	Hepatoma cells	HCC progression
	miR-34a [170]	Yes	Hepatocytes and mouse tumors and patients with HCC	Liver tumorigenesis
	miR-34a [164]	Yes	Mouse liver tissues and hepatic cell lines	Lipid and lipoprotein metabolism in NAFLD
	H19 [163]	N.A.	Hepatic cells and mouse fetal livers	Gluconeogenesis
	miR-21 [166]	Yes	HSCs and hepatocytes	Liver fibrosis
	let-7b [171]	No	Human adipose tissue-derived mesenchymal stem cells	Hepatic differentiation
RXRα	miR-27a and -27b [183]	Yes	Rat primary HSCs	Lipid metabolism and cell proliferation in HSC activation
	miR-34a [184]	Yes	Hepatic cell lines and human liver tissues	Drug- or lipid-metabolism and liver fibrosis
	miRNA-128-2 [185]	Yes	Hepatic cell lines	Cholesterol homeostasis
ERα	miR-939-3p [193]	Yes	HCC cell lines and tissues	HCC progression
	miR-18a [194]	Yes	HCC patient livers and cell lines	Development of HCC
	miR-22 [195]	Yes	HCC patient livers and cell lines	Development of HCC
	miR-221 [196]	Yes	HCC cell lines	HBV-related HCC
	miR-26a [197]	Yes	Liver cancer cell lines, mouse xenograft model, and patient livers	HCC progression
ERRα	SPRY4-IT1 [201]	N.A.	HCC tissues and cell lines	Malignancy and progression of HCC
ERRγ	miR-940 [202]	Yes	HCC tissues and cell lines	Cell proliferation and apoptosis in HCC
	miR-545 [203]	Yes	HCC tissues and cell lines	HCC prognosis
GR	miR-124a [206, 207]	Yes	Liver tissues and hepatic cells	Lipid metabolism
	GAS5 [39, 208]	N.A.	Hepatic and other cell lines	Autoimmune and inflammatory response
MR	miR-766 [210]	Yes	Cell lines and xenograft model	HCC progression
LRH1	miR-10a [216]	Yes	Human liver tissues and cell lines	HCV-related cirrhosis
	miR-1275 [217]	Yes	Porcine granulosa cells	Apoptosis and estrogen secretion
	miR-381 [218]	Yes	Human HCC cell lines and patient tissues	HCC progression
DAX-1	miR-561 [220]	Yes	HepG2 cells and human primary hepatocytes	Acetaminophen-induced hepatotoxicity
SHP	miR-142-3p [222]	Yes	Cholic acid-fed mice	Cholestasis
	MEG3 [223]	N.A.	Mouse liver tissues and hepatic cells	Cholestatic liver injury
	H19 [225]	N.A.	Hepatic cells and exosomes	Cholestatic liver injury

2. Crosstalk between NRs and ncRNAs in liver pathophysiology

2.1. Molecular function of NRs and ncRNAs

The emerging concept in the field of NRs is the NR crosstalk that refers to the interplay between different NRs or between their overlapping signaling pathways [22]. NRs act in the form of “typical” heterodimers (e.g., retinoid X receptor (RXR) partnering with PPARs or retinoic acid receptors (RARs)) or “atypical” heterodimers (e.g., PPARs partnering with estrogen-related receptors (ERRs)). The latter function more flexibly and transiently and do not strictly need DNA binding of both partners [23, 24]. The obligatory dimerization partners are often co-expressed in specific tissues and share the similarity in domains; thus, one ncRNA might be able to target multiple NRs to mediate the crosstalk between different NR signaling pathways.

NRs regulate the transcription of ncRNAs as direct TFs (Figure 2A). The positive or negative regulation of specific ncRNAs by NRs is discussed below in individual sections. The expression of ncRNAs is closely connected to the promoter activity, chromatin conformation, and DNA elements of their transcription loci. NRs directly or indirectly bind to the promoters of miRNA host genes to regulate transcription. A miRNA residing in the intron or exon of a protein-coding gene can share its host gene’s promoter; however, a miRNA gene often has multiple transcription start sites. The promoter of an intronic miRNA can be distinct from that of its host gene [25]. A genome-wide comparative analysis of the promoters of human IncRNAs and protein-coding genes shows the global differences in specific genetic and epigenetic features relevant to the transcriptional regulation between these two groups of promoters, suggesting that distinct TFs significantly favor IncRNA, rather than coding-gene, promoters [26]. Active enhancers are a source of functional ncRNAs, generally producing unspliced and non-polyadenylated transcripts termed enhancer RNAs (eRNAs). These enhancers are characterized by NR-dependent recruitment of mega-dalton-sized multi-transcription factor complex and high eRNA production [27–29]. The precise binding sites of NRs on ncRNA genes can be inferred from collective analysis of CpG islands and chromatin immunoprecipitation followed by sequencing (ChIP-seq) data [30]. For instance, DIANA-miRGen indexes interactions between TFs and miRNAs based on the TF motifs for more than 1,500 miRNAs [31], and FANTOM (functional annotation of the mammalian genome) provides atlases of mammalian promoters, enhancers, IncRNAs, and miRNAs [32].

Figure 2.

The molecular function of NRs and the crosstalk between NRs and ncRNAs. (A) NRs regulate the expression of ncRNAs. NRs bind to their responsive DNA elements and wherein activate or repress the expression of ncRNA host genes via association with transcriptional cofactors. (B and C) The action mechanisms of miRNAs. MiRNAs bind to the complementary sequences (seed sequences) in the 3′-UTR of target mRNAs, leading to gene expression silence via mRNA decay (B) or translational inhibition (C). The mature miRNA is part of an active RNA-induced silencing complex (RISC) containing Argonaute (AGO) proteins which cleave the mRNA and lead to direct mRNA degradation. (D) The action modes of lncRNAs: (a) lncRNAs serve as scaffolds in the formation of ribonucleoprotein (RNP) complex which can epigenetically modify chromatin structure; (b) lncRNAs transcribed from enhancer regions (eRNAs) can contribute to chromatin looping and gene activation; (c) guide lncRNAs activate or repress gene expression through recruiting or evicting regulatory cofactors; (d) decoy lncRNAs complete with DNA response elements to inhibit regulatory protein factors bound to the genome, thus terminating gene transcription; (e) lncRNAs function as molecular sponges and compete for miRNA binding to mRNAs, thus inhibiting miRNA-mediated gene repression; (f) spatiotemporally-expressed specific IncRNAs can be specific signals or indicators of certain biological events; (g) lncRNAs participate in RNA splicing; (h) lncRNAs modulate gene translation and mRNA stability; (i) lncRNAs can be the primary transcripts of miRNAs.

MiRNAs are highly conserved between species, with approximately 22 nucleotides in length. MiRNAs negatively regulate gene expression, primarily but not exclusively, by binding to microRNA response elements (MREs) within the 3′-untranslated region (3′-UTR) or the coding region of target mRNAs, resulting in degradation or translational repression of gene targets (Figure 2B and 2C) [33]. A miRNA can simultaneously regulate the expression of multiple target mRNAs, leading to the coordination of different biological pathways. A specific miRNA can also bind to multiple sites within a target mRNA, improving the efficacy to control target gene expression. MiRNAs have important implications in liver pathophysiology [34–36].

LncRNAs can silence gene expression via modulation of translation and mRNA stability. However, additional distinct action mechanisms for IncRNAs have been revealed, such as participating histone modification to remodel chromatin, directing the recruitment of RNA polymerase and cofactors, and involving in alternative splicing, etc. (Figure 2D) [21, 37, 38]. LncRNAs can act as decoys to prevent NRs from binding to their DNA elements, e.g., growth arrest-specific 5 (GAS5) [39]. LncRNAs can serve as scaffolds and associates with NRs and other proteins to mediate transactivation, e.g., steroid receptor RNA activator (SRA) [40]. About 20% of IncRNAs are eRNAs and participate in chromosomal enhancer-promoter looping [41]. LncRNAs can cooperate with miRNAs to regulate gene expression. For instance, IncRNA-ATB competes with miR-425-5p to up-regulate TGF-β type II receptor (TGF-βII) and SMAD2 to promote liver fibrosis [42]; IncRNA-ATB also acts as a sponge of miR-200a to maintain β-catenin expression [43]. Besides, IncRNAs can be the precursors of miRNAs and serve as signals for the activation of specific biologic events [38, 44].

2.2. Thyroid hormone receptors (TRs) and ncRNAs

TRs (TRβ and TRβ) are ligand-dependent TFs that mediate the function of thyroid hormones (THs), 3,5,3′-L-triiodothyronine (T3) and L-3,5,3′,5′-tetraiodothyronine (T4). TRs dimerize with different NRs to regulate the expression of different genes. The TR/RXR dimer displays the highest transcriptional activity.

The TH/TR signaling has a prominent impact on fatty acid and cholesterol metabolism in the liver. Inhibition of miR-383 and miR-146b expression increases TRβ expression and decreases lipid accumulation in fatty acid-treated mouse primary hepatocytes [45, 46]. THs increase miRNA-181d expression and decrease cholesterol secretion through the inhibition of its gene targets, homeobox transcription factor 2 (CDX2) and O-acyltransferase 2 (SOAT2), in HepG2 cells and mouse livers [47].

TRs are implicated in tumorigenesis. However, the role of TRs in hepatocellular carcinoma (HCC) is still elusive and might depend on the receptor subtype and tumor stage [48]. The isoform TRβ1 is down-regulated in early pre-neoplastic lesions and HCC in rat livers, which coincides with the induction of three TRβ1 targeting miRNAs, miR-27a, -181a, and -204 [49]. This suggests that a hypothyroid state might contribute to the progression from pre-neoplastic lesions to HCC. In agreement with this tumor-suppressive role, THs induce miR-214-3p expression and inhibits HCC formation [50]. In contrast, T3/TRs stimulates the expression of miR-21 and miR-17 to enhance migration and invasion of hepatoma cells [51, 52]. T3/TRs also negatively regulates miR-130b expression to enhance the motility and invasion of hepatoma cells [53].

2.3. RARs and ncRNAs

RARs (RARα, RARβ, and RARγ) are activated by both all-trans retinoic acid and 9-cis retinoic acid. In the absence of a ligand, RAR dimerizes with RXR and binds to retinoic acid response elements (RAREs), complexed with corepressor proteins. The binding of an agonist to RAR leads to the dissociation of corepressors and the recruitment of coactivators, which activates transcription of genes implicated in differentiation, growth, apoptosis, inflammation, and immune responses [54].

RARβ serves as a tumor suppressor whose promoter is regulated by methylation. Polyinosinic:polycytidylic acid (poly I:C)-mediated activation of TLR3 induces the expression of several miRNAs (miR-29b, -29c, -148b, and -152) that target DNA methyltransferases, leading to gene demethylation, RARβ re-expression in a variety of cancer cells, and tumor regression [55]. Although dozens of miRNAs are predicted to target RARs, few studies show that RAR transcripts are directly targeted by miRNAs in liver cells [56, 57], and RARs crosstalk with ncRNAs in liver disease is far less explored than RXRs [58]. It is noteworthy that the ncRNA regulation of NRs relies on tissue specificity. For instance, several miRNAs, including miR-9, are predicted to target RARα, RARβ, RXRα, and RXRβ in silico, but not responsible for the down-regulation of these NRs in pancreatic duct adenocarcinoma [59].

2.4. PPARs and ncRNAs

2.4.1. NcRNA regulation of PPARα

2.4.1.1. Regulation of PPARα by miRNAs

PPARs (PPARα, PPARβ/δ, and PPARγ) heterodimerize with RXRs to regulate gene transcription. PPARα is the predominant PPAR isoform in the liver and is mainly expressed in liver parenchymal cells. PPARα is involved in fatty acid metabolism, lipid export, hepatic glucose production, and ketone body synthesis [60].

PPARα is the target of several miRNAs such as miR-21 [61, 62], miR-17-5p [63], miR-27b [64], miR-34a [65], and miR-540 [66]. The abnormal crosstalk between these miRNAs and PPARα leads to the dysregulation of lipid metabolism and implicates in NAFLD/NASH pathogenesis [61–66].

Inositol-requiring enzyme 1α (IRE1α) is critical in maintaining lipid homeostasis in the liver by repressing miRNA biogenesis. IRE1α is inactivated in mice fed a high-fat diet and patients with steatohepatitis, which up-regulates the expression of miR-200 family and miR-34a, leading to the decreased expression of their targets, PPARα and SIRT1 [67]. MiR-10b enhances cellular steatosis by targeting PPARα in cultured hepatocytes [68]; a reverse correlation between miR-10b and PPARα is also revealed in NAFLD patients [69]. PPARα regulates alcohol-associated steatosis, and its expression is impaired in mouse models of alcohol-associated liver disease (ALD) [70]. The decrease of PPARα mRNA in alcohol-fed mice is prevented by miR-155 knockout, accompanied by the increase of PPARα binding to PPAR response element and the attenuation of chronic alcohol-induced steatosis and liver injury [70].

In a functional high-throughput screening, miR-9 is identified to target PPARα in human HCC cells; a miR-9/PPARα/CDH1 signaling pathway is revealed in HCC oncogenesis [71]. The miR-9/PPARα axis is also validated to modulate lipid metabolism in hepatoma cells [72].

2.4.1.2. Regulation of PPARα by IncRNAs

A circular RNA, circRNA_0046366, is recognized as a specific antagonist of miR-34a. The circRNA_0046366/miR-34a/PPARα axis represents a novel epigenetic mechanism underlying the occurrence and resolution of hepatic steatosis [73]. LncRNA HULC up-regulates PPARα and modulates lipid metabolism in HCC cells by activating the acyl-CoA synthetase subunit ACSL1. HULC induces methylation of CpG islands in the miR-9 promoter to inhibit miR-9 expression, thus promoting PPARα expression [72]. In microarray analyses, 3,360 IncRNAs (2,048 up-regulated and 1,312 down-regulated) are differentially expressed in livers of diabetic db/db mice with NAFLD and the IncRNA AK012226 is identified as a potential regulator of PPAR signaling pathway, although the direct experimental evidence is lacking [74]. The IncRNA XLOC_009593 is also predicted to regulate PPARα in minipigs with NASH [75].

2.4.2. NcRNA regulation of PPARγ

PPARγ is abundantly expressed in adipose tissue, regulating adipocyte differentiation, adipogenesis, and lipid metabolism. Hepatic PPARγ expression is robustly increased in NAFLD patients and experimental models. Hepatocyte- and macrophage-specific PPARγ knockout alleviates high-fat diet-induced hepatic steatosis in mice [76]. MiR-27b targets PPARγ 3’-UTR [77] and contributes to lipopolysaccharide-mediated PPARγ mRNA destabilization [78]. When the endogenous miR-27b activity is disrupted by a transgenic miR-27b sponge in zebrafish, lipid accumulation and the expression of the adipogenic transcription factor PPARγ is enhanced in the liver, with an early onset of NAFLD and NASH [79]. In the livers of male fetuses of gestational diabetes mellitus (GDM) rats, a miR-130/PPARβ axis is dysregulated; instead of that, a miR-9/PPARβ/δ axis is dysregulated in the female, suggesting sex-dependent alteration of miRNA/PPAR axes in lipid metabolism in fetal livers of GDM rats [80].

PPARβ suppresses hepatic stellate cell (HSC) differentiation. Loss of PPARβ renders HSC myofibroblastic activation, a key event in liver fibrosis [81]. PPARγ is a direct target of miR-34a/34c [82], miR-130a/130b [83], and miR-942 [84]; they repress PPARγ expression via directly binding to the 3’-UTR of PPARγ mRNA and contribute to HSC activation in humans or rats. As an anti-fibrotic gene, PPARγ is also targeted by miR-155 during the development of ALD and carbon tetrachloride (CCl4)-induced liver fibrosis in mice [70]. MiR-29a indirectly increases PPARγ) expression via an epigenetic mechanism of inhibiting bromodomain-containing protein 4 (BRD4) and enhancer of zeste homolog 2 (EZH2), thus repressing HSC activation and mitigating live fibrosis [85]. Interestingly, extracellular vesicles (EVs) shuttle miR-128-3p, a PPARγ-targeting miRNA, from fat-laden hepatocytes into HSCs to reduce PPAR γlevels and induce HSC activation [86].

PPARγ counteracts the occurrence and progression of liver cancer [87]. Activation of PPARγ triggers apoptosis and inhibits cell growth and metastasis in HCC [88]. PPARγ is directly targeted by miR-130b, a miRNA promoting epithelial-to-mesenchymal transition (EMT) and invasion in HCC cells [89]. PPARγ is also targeted by miR-27a, a miRNA that is highly enriched in both serum and liver cancer tissues of patients and promotes HCC cell proliferation [90, 91].

2.4.3. PPAR regulation of ncRNAs

PPAR-dependent induction of miRNAs and several miRNAs regulating PPARs in specific tissues and pathophysiological processes have been reviewed previously [92]. In the liver, activation of PPARα using an agonist WY-14643 inhibits let-7c expression, leading to increased hepatocellular proliferation and tumorigenesis [93]. Inhibition of PPARα using an antagonist, 2-chloro-5-nitro-N-(pyridyl)benzamide, represses the expression of miR-7 which may serve as a crosstalk coordinator between the PPARα, liver X receptor β (l_XRβ), and sterol regulatory element-binding protein (SREBP) signaling pathways [94].

PPARγ activates miR-122 expression in GDM rat livers [95] and HCC cells [96]. 1, 4-bis [2-(3,5-dichloropyridyloxy)] benzene (TCPOBOP), a potent chemical mitogen for the liver and an agonist of CAR, reduces miR-122 expression, which can be inhibited by PPARγ agonist rosiglitazone [97]. A panel of antifibrotic miRNAs, including let-7, miR-30, miR-29c, miR-335, and miR-338, collectively termed antifibrotic miRNAs (AF-miRNAs), is transcriptionally and positively regulated by PPARγ in mouse liver fibrosis and HCC models [98].

2.5. REV-ERBs, retinoic acid receptor-related orphan receptors (RORs), and ncRNAs

REV-ERBs, including REV-ERBα (NR1D1) and REV-ERBβ (NR1D2), have been deorphanized with heme identified as their natural ligands. RORs, including RORα (NR1F1), RORβ (NR1F2), and RORγ (NR1F3), display significant sequence similarities and functional overlap and are differentially expressed in different tissues; each member has multiple isoforms resulted from alternative promoter usage and exon splicing at the amino-terminal region of the receptor [99]. Endogenous ligands for RORs include cholesterol and its derivatives. All RORs recognize and bind as monomers to ROR response elements (ROREs) and are often co-expressed in the same tissue as REV-ERBs. The ROREs are often shared by REV-ERBs that bind as a monomer or homodimers and act as transcriptional repressors due to lacking the activation-function 2 (AF-2) region. Both REV-ERBs and RORs are critical circadian NRs that regulate circadian rhythms, metabolism, and lipid homeostasis [100]. Their pharmacological implications have been reviewed elsewhere [101].

Mice exposed to artificial light at night develop obesity and hepatic steatosis, with an altered expression profile of miRNAs targeting REV-ERBα, including miR-140-5p, −185-5p, -326-5p, and -328-5p, indicating that miRNAs are important regulators in maintaining lipid homeostasis by tuning the expression of circadian NRs and highlighting the concept that disruption of the peripheral clock can lead to hepatic steatosis [102]. In an extensive analysis of miRNA-mRNA co-expression in circadian rhythms, miR-324-3p is predicted to target REV-ERBα [103]. RORα is validated to be the target of miR-1246. Either overexpression of miR-1246 or knockdown of RORα enhances the proliferation, invasion, and metastasis of HCC cells [104].

Multiple miRNAs are rhythmically transcribed in mouse livers and regulated by key circadian transcription factors such as BMAL1/CLOCK and REV-ERBα/β [105]. REV-ERBα is the major circadian regulator of miR-122 that is a highly abundant and hepatocyte-specific miRNA and contributes to circadian control through regulation of many circadian genes, including PPARβ/δ and the PPARα coactivator SMARCD1/BAF60A. This suggests that miR-122 is an important mediator for the crosstalk between circadian NRs [106]. Besides, BMAL1 and REV-ERBaαbind to the enhancer foci in the murine genome, where a significant proportion of circadian IncRNAs is expressed, demonstrating circadian regulation of IncRNAs [107].

2.6. Liver X receptors (LXRs) and ncRNAs

LXRs (LXRα and LXRβ) are master regulators in the metabolism of cholesterol, lipids, bile acids, and steroid hormones [108]. Ligand-activated LXRs form heterodimers with RXRs and regulate the expression of genes containing LXR response elements, such as the lipogenic gene sterol regulatory element-binding transcription factor 1 (SREBF1, also known as SREBP-1) and the cholesterol transporter gene ATP binding cassette subfamily A member 1 (ABCA1). LXRα expression is subjected to autoregulation because LXR ligands induce LXRα expression and LXRα can regulate its own promoter activity [109].

LXRs’ target gene SREBP-1c (SREBF1 isoform 1c) plays a crucial role in inhibiting HSC activation. MiR-27a/b-3p reduces LXRα and SREBP-1c levels independently of their 3′-UTRs, thus increases HSC activation and liver fibrosis [110]. MiR-7 mediates the interplay between PPARα and LXRβ in hepatic lipid metabolism, as LXRβ is a direct target of miR-7 that can be induced by PPARα activation [94]. Interestingly, pharmacological activation of LXRs promotes miR-7 expression in neuroblastoma cells, suggesting that miR-7 and the LXR signaling might operate in a negative-feedback loop fashion [111].

The negative feedback between NRs and miRNAs is observed in the regulation of cholesterol and lipid homeostasis. MiR-613 targets LXRα, while LXRα positively regulates miR-613 expression by inducing SREBP-1c expression [112]. MiR-613 suppresses the expression of both LXRα and ABCA1 (transferring excess cellular cholesterol to lipid-poor apolipoproteins) by targeting their 3′-UTRs and inhibits cholesterol efflux in PPARγ activated THP-1 macrophages [113]. Except for the role in cholesterol efflux, miR-613 also inhibits lipogenesis in HepG2 cells by down-regulating LXRα [114].

LXRα is the target gene of miR-155 that is down-regulated in the liver and peripheral blood of NAFLD patients and can reduce hepatic lipid accumulation by suppressing LXRα-dependent lipogenic signaling pathway [115]. MiR-1 and miR-206 also attenuate LXRα-induced lipogenesis by targeting LXRα 3′-UTR and repress LXRα-induced lipid droplet accumulation in hepatocytes [116]. A negative feedback loop between miR-206 and LXRα is intriguingly shown in macrophages. MiR-206 overexpression up-regulates LXRα to enhance cholesterol efflux in THP-1 cells; on the other hand, miR-206 expression in macrophages can be repressed by LXRα activation [117]. The miR-33 family, including miR-33a and miR-33b, are well-characterized for their roles in regulating lipid metabolism and cholesterol trafficking and efflux [118]. The intronic hsa-miR-33a and hsa-miR-33b are located within and co-transcribed with SREBF-2 and SREBF-1, respectively. MiR-33a and miR-33b have not been validated to target LXRs, but miR-33 inhibition impairs LXR activation (by agonist T0901317)-induced LXRα elevation and decreases the lipogenic effects of LXRα in mouse livers [119].

LXRα is a transcriptional activator of miR-378. The LXRα agonist GW3965 activates miR-378 transcription and reduces peroxisome proliferator-activated receptor γ coactivator 1-beta (Ppargc1β) expression, which subsequently impairs fatty acid oxidation and aggravates hepatic steatosis [120]. LXRα activation by GW3965 also increases the expression of hepatic IncRNA CHROME, promoting cholesterol efflux and nascent high-density lipoprotein (HDL) particle formation. CHROME knockdown in human hepatocytes and macrophages increases the expression of miR-27b, miR-33a, miR-33b, and miR-128, leading to the disruption of miRNA-mediated gene networks and related biologic functions [121]. This LXRα-CHROME-miRNAs network is a central component of the ncRNA circuitry controlling cholesterol homeostasis in humans. Besides, LXR activation inhibits miR-26 and increases miR-144 expression in macrophages and mouse livers [122, 123].

2.7. FXR and ncRNAs

FXR is the bile acid receptor and highly expressed in the liver and intestine. When activated by bile acids, FXR heterodimerizes with RXR to inhibit bile acid synthesis through the induction of small heterodimer partner (SHP) to suppress the expression of cholesterol 7 alpha-hydroxylase (CYP7A1) and sterol 12-alpha-hydroxylase (CYP8B1), two rate-limiting enzymes in bile acid biosynthesis from cholesterol [124, 125]. FXR plays pleiotropic roles in bile acid biosynthesis, lipid metabolism, glucose homeostasis, inflammation, and carcinogenesis.

2.7.1. NcRNA regulation of FXR

NcRNAs regulate FXR expression in lipid metabolism. For instance, miR-194 directly inhibits FXR expression, and miR-194 inhibitor increases FXR expression in livers of obese mice and improves obesity-induced NAFLD and metabolic disorders [126]. A YY1-FXR-SHP signaling is proposed to be critical to lipid homeostasis. MiR-122 inhibits lipid droplet formation and hepatic triglycerides accumulation in NASH/NAFLD mice via targeting the transcriptional repressor YY1 (a driving force in hepatic steatosis), which results in the up-regulation of FXR and SHP [127]. A liver-enriched lncRNA, liver-specific triglyceride regulator (lncLSTR), maintains systemic lipid homeostasis by regulating a TDP-43/FXR/apoC2-dependent pathway in mice [11].

Loss of FXR promotes carcinogenesis in the liver [128]. FXR is targeted by the oncogenic miR-421 that promotes cell proliferation, colony formation, migration, and invasion in biliary tract cancer (BTC) [129] and HCC cells [130]. However, FXR is also target by miR-192 that suppresses proliferation in hepatic Huh-7 cells [131].

2.7.2. FXR regulation of ncRNAs

FXR regulates miRNA expression in the pathophysiology of liver disease. FXR activates miR-29a to negatively regulate the expression of extracellular matrix (ECM)-producing genes in HSCs during liver fibrosis [132]. Activation of FXR by an agonist GW4064 increases hepatic levels of miR-144 and decreases plasma HDL-cholesterol levels [133]. FXR is a transcriptional activator of miR-22, and FXR knockout in mice decreases hepatic miR-22 expression and potentiates hepatic cell proliferation [134]. FXR also up-regulates miR-122 levels to suppress the proliferation of HCC cells in vitro and the growth of HCC xenografts in vivo [135]. Further, FXR promotes miR-126a expression in cultured hepatic cell lines [136]. Treatment of mice with FXR-specific agonists (GSK2324 or GW4064) rapidly increases the hepatic level of miR-33 [137]. On the other hand, FXR indirectly inhibits hepatic miR-34a expression through SHP-mediated inhibition of p53 activity in obese mice [138]. FXR inhibits miR-186 expression to attenuate ER stress-induced hepatocyte death and liver injury [139]. FXR also protects hepatocytes from oxidative stress-induced injury by repressing miR-199a-3p expression [140].

2.8. Vitamin D receptor (VDR) and ncRNAs

VDR (NR1I1) is widely expressed in most tissues. It forms a heterodimer with RXR to bind DNA response elements on target genes. A variety of functions of VDR have been revealed through the investigation of its ligand vitamin D.

Deficiency of vitamin D contributes to autoimmunity, and defects of VDR signaling are associated with the pathogenesis of autoimmune liver diseases, such as primary biliary cholangitis (PBC) and primary sclerosing cholangitis (PSC), two cholestatic liver disorders. VDR expression is substantially suppressed in the livers of patients with PBC. Insufficiency of VDR signaling enhances miR-155 expression to decrease the expression levels of SOCS1, a miR-155’s target gene suppressing JAK/STAT pathway and inflammatory cytokine production [141]. Since miR-155 targets PPARα, PPARγ, and LXRα, it is possible that miR-155 can mediate the crosstalk between VDR and PPAR or LXR signaling pathways.

VDR is a target of miR-125a-5p that inhibits astemizole-induced VDR expression and impairs the synergistic anti-tumor effect of astemizole and vitamin D in HCC [142]. VDR is also targeted by miR-351 that promotes schistosomiasis-induced liver fibrosis and inhibits VDR’s anti-fibrotic effect of suppressing SMAD-mediated activation of fibrotic genes in HSCs [143].

2.9. Pregnane X receptor (PXR) and ncRNAs

PXR (NR1I2) is the major regulator of xenobiotic metabolism by transcribing a variety of transporters and drug-metabolizing enzymes, including the cytochrome P450 family [144]. In addition to its canonical xenobiotic sensing function, PXR participates in glucose, lipid, and bile acid metabolism, energy homeostasis, inflammatory response, cell proliferation, apoptosis, and cell migration [145].

MiR-148a targets PXR and PXR’s target gene CYP3A4. The translational efficiency of PXR (PXR protein/PXR mRNA ratio) is inversely correlated with the level of miR-148a in human livers (pathology was not shown) [146]. MiR-148a/PXR axis is also involved in estrogen-induced cholestasis during pregnancy [147].

PXR is directly targeted by miR-18a-5p, which contributes to the down-regulation of PXR by its ligand rifampicin and the up-regulation of PXR by glucocorticoids in hepatic cells [144]. LINC00844, a sponge of hsa-miR-486-5p, positively regulates the expression of PXR, hepatocyte nuclear factor 4α (HNF4α), and a group of drug-metabolizing enzymes including CYP3A4, SULT2A1, and CYP2E1 in response to the toxicity of N-acetyl-para-aminophenol (APAP) in HepaRG cells and human primary hepatocytes [148]. The IncRNA hepatocyte nuclear factor 1 alpha antisense 1 (HNF1α-AS1), an antisense RNA of HNF1α, positively regulates PXR expression in hepatic Huh7 cells [149].

Using PXR agonist pregnenolone-16α-carbonitrile (PCN) and CAR agonist TCPOBOP, PXR- and CAR-regulated IncRNAs have been profiled in mouse livers, and 774 IncRNAs genes with direct PXR-DNA binding sites are identified [150].

2.10. CAR and ncRNAs

CAR belongs to the vitamin D receptor-like subfamily of NRs, along with VDR and PXR. CAR functions as a xeno-sensor and regulates xenobiotic detoxification, lipid homeostasis, energy metabolism.

The role of CAR in hepatocarcinogenesis and drug metabolism has been extensively studied using its specific agonist TCPOBOP [151]. CAR is well recognized as an ADME (Absorption, Distribution, Metabolism, and Excretion) gene which regulates both the expression of other ADME genes (including AHR, ARNT, HNF4A, PXR, PPARA, PPARD, PPARG, and RXRA) [152] and the levels of ncRNAs.

CAR activators, phenobarbital (PB) and chlordane, induce the expression of ncRNAs from Dlk1-Dio3 locus, which is an early biomarker for CAR activator-induced nongenotoxic hepatocarcinogenesis [153, 154]. CAR activation by TCPOBOP dysregulates the expression of IncRNAs proximal to CAR target genes (Cyp2b10, Por, and Alas1) in mouse livers. TCPOBOP also decreases the level of miR-122 by suppressing the transcriptional activity of HNF4α in mouse livers [155]. MiR-122 is down-regulated by PB, contributing to AMPK-dependent CAR activation and the induction of CAR’s target genes [156, 157]. In a miRNA array profiling, a panel of mouse miRNAs is identified to be regulated by either PCN or TCPOBOP. Among them, 9 Cyp genes are identified as candidate targets of PCN-regulated miRNAs, while16 Cyp genes are identified as candidate targets of TCPOBOP-regulated miRNAs, suggesting that miRNAs can modulate the expression of P450 enzymes driven by PXR/CAR activation in the liver [158]. Fifty-one miRNAs, identified by RNA-sequencing, are significantly altered by TCPOBOP in mouse livers, including miR-802-5p and miR-485-3p [159].

CAR increases the expression of the bilirubin-clearance pathway components to mediate PB’s hepatic effect on bilirubin clearance [160]. MiR-137 is a repressor of CAR expression and thus serves as a critical regulator of bilirubin clearance in neonatal jaundice [161].

2.11. HNF4α and ncRNAs

HNF4α (NR2A1) has the highest expression in the liver compared to other organs. HNF4α binds to the promoter of 12% of genes expressed in the adult liver and regulates the expression of many genes involved in the metabolism of bile acids, lipids, glucose, amino acids, and xenobiotics. HNF4α participates in liver development during gastrulation and directs differentiation of hepatocytes from hepatoblasts [162].

2.11.1. ncRNAs regulation of HNF4α

As a metabolic regulator, HNF4α regulates the expression of rate-limiting enzymes in gluconeogenesis. The IncRNA H19 decreases DNA methylation and increases Hnf4α expression, leading to concomitant activation of the gluconeogenic program in hepatic cells, which might contribute to the fetal origin of adult metabolic abnormalities [163]. Besides, the miR-34a-HNF4α pathway is activated by metabolic stress in both NASH patients and diabetic or high-fat diet-fed mice, with remarkable inhibition of HNF4α expression; miR-34a suppresses very-low-density lipoprotein secretion and promotes hepatic steatosis and hypolipidemia in a HNF4α-dependent manner [164].

As a fibrosis regulator, HNF4α deactivates myofibroblasts through inducing mesenchymal-to-epithelial transition (MET) and suppressing their proliferation [165]. HNF4α is an indispensable and dominant regulator of the epithelial phenotype in liver cells. MiR-21 directly targets HNF4α and modulates ERK1 signaling and EMT during liver fibrosis [166].

As a tumor suppressor of HCC, HNF4α is a potent inducer of MET and a promising target for gene therapy [167]. Transient inhibition of HNF4α initiates hepatocellular oncogenesis through a miRNA-involved inflammatory feedback loop consisting of miR-124, IL6R, STAT3, miR-24, and miR-629 [168] MiR-24 and miR-629 suppress HNF4α expression, thus inducing hepatocellular transformation; HNF4α activates miR-124 expression, while miR-124 inactivates the inflammatory IL-6R/STAT3 pathway through direct inhibition of IL-6R expression; STAT3, in turn, binds the promoters of miR-24 and miR-629 to activate their expression [168]. This miR-24/miR-629/HNF4α/miR-124/STAT3 loop argues for an epigenetic alteration as the major event in cancer initiation. It supports the concept that the use of miRNA mimics or inhibitors can be a potential therapeutic approach in the prevention of liver cancer. A feedback loop consisting of HNF4α, NF-κB, miR-7, miR-124, and miR-21 is also identified during HCC progression. MiR-7 and miR-124 are transcriptionally activated by HNF4α, while miR-21 directly inhibits HNF4α levels through targeting HNF4α 3′-UTR [169]. Using the locked nucleic acid (LNA)-34a (an inhibitor of miR-34a, consisting of the complementary sequence of miR-34a), a study shows that miR-34a inhibits HNF4α expression in the liver. LNA-34a can hamper the progression of liver cancer harboring β-catenin mutations [170].

In line with the commitment of HNF4α in hepatocyte differentiation, inhibition of let-7b can activate the hepatic differentiation of human adipose tissue-derived mesenchymal stem cells (hAT-MSCs) through indirectly up-regulating HNF4a expression [171]. Interestingly, both gain- and loss-of-function of miR-337-3p inhibit HNF4α expression through a biphasic incoherent feedforward loop in liver development, but the regulation is not realized through the interaction between miR-337-3p and HNF4α 3′-UTR [172].

2.11.2. HNF4α regulation of ncRNAs

HNF4α is a transcriptional activator of miR-122 in mouse livers and HCC cell lines [173, 174]. This HNF4α/miR-122 pathway is associated with iron overload-mediated hepatic inflammation [175]. HBV-associated hepatocarcinogenesis [176], and gluconeogenic and lipid metabolic alterations in type 2-diabetic mice and palmitate-treated HepG2 cells [177]. JNK1-mediated inactivating phosphorylation of HNF4α is responsible for the transcriptional inhibition of miR-122, which leads to hepatic insulin resistance [178]. In a microarray profiling of miRNAs in rat neonatal liver-derived epithelial cells transduced by lentiviral HNF4α, 26 miRNAs are significantly altered, but not miR-122 [179]. In hepatocyte-specific Hnf4α knockout mice, it is demonstrated that HNF4α is essential for hepatic basal expression of a group of liver-enriched miRNAs, including miR-101, miR-192, miR-193a, miR-194, and miR-802 [180]. The transactivation of miR-194/192 by HNF4α is also confirmed by another study that suggests that the HNF4α-miR-194/192 signaling axis may play a critical role in maintaining the differentiation status of hepatocytes [181]. HNF4α also directly binds promoters of miR-15/16 and promotes their transcription during the differentiation of amniotic epithelial cells into hepatocytes [182].

2.12. RXRs and ncRNAs

RXRs (RXRα, RXRβ, and RXRγ) are activated by retinoic acids and heterodimerize with other NRs, including CAR, FXR, LXRs, PPARs, PXR, RARs, TRs, and VDR [24].

RXRα plays a central role in adipogenesis and HSC proliferation. MiR-27a and −27b target RXRα to promote lipid catabolism and cell proliferation during HSC activation [183]. RXRα positively regulates the expression of drug- or lipid-metabolizing enzymes such as CYP26. MiR-34a can decrease CYP26 expression by targeting RXRα in HepG2 and human fibrotic livers [184]. RXRs sense cellular cholesterol concentration, mediate cholesterol efflux, and maintain cellular sterol homeostasis, as evidenced by that RXRs cooperate with LXRs to induce the transcription of key cholesterol transporters, ABCA1 and ABCG1, as well as that of apolipoprotein E. MiR-128-2 is identified as a regulator in cholesterol homeostasis through curtailing the expression of ABCA1, ABCG1, and RXRα [185].

The IncRNA HULC is increased and aids in lipid dynamics during hepatitis virus C (HCV) infection, as a result of RXRα-mediated transcription [186]. That RXR induces ncRNA expression is also demonstrated in other cell types. For instance, the activation of RXRα/ RARα induces miR-10a expression in vascular endothelial cells [187].

2.13. Estrogen receptors (ERs) and ncRNAs

The liver is a hormone-responsive and sexually dimorphic organ that expresses androgen receptor (AR) and estrogen receptor ERα (also known as ESR1, the dominant ER type in hepatocytes). It displays disparity in the responsiveness to sex hormones in terms of gene expression patterns, immune responses, and xenobiotic metabolism. The importance of sex hormone receptors in liver pathophysiology is not well-understood.

The dysregulation of ER/miRNA network is associated with NAFLD [188]. For instance, estrogen can activate miR-125b expression to protect against hepatic steatosis induced by a high-fat diet in mice [189]. The IncRNA NEAT1.1 can interact with ERα and might regulate steatosis in HepG2 cells [190].

Loss of ER signaling is closely related to the development of HCC. ERα protein is down-regulated in 60% of female HCC cases; the estrogen pathway shows a tumor-protective function in female hepatocarcinogenesis [191], which might be related to its anti-inflammation function [192]. ERα is a target of miR-939-3p that is up-regulated in HCC cell lines and tissues [193]. ERα is also a target of miR-18a that promotes HCC cell proliferation [194]. ERα is targeted by miR-22 that is highly expressed in male HCC tumor-adjacent tissues and potentially promotes HBV-related HCC development in males [195]. ERα is also targeted by miR-221 which is highly expressed in HCC cells transfected with Hepatitis B virus X protein (HBx), an important protein promoting cell proliferation and growth in the development of HCC [196]. Interestingly, ERα is also targeted by the tumor suppressor miR-26a, a miRNA reduced in HCC and inhibiting human liver cancer cell growth in vitro and in a xenograft model [197].

Estrogen transcriptionally activates the expression of miR-23a that targets the anti-apoptotic gene XIAP [198]. Estrogen inhibits the expression of miR-21, an onco-miR [199]. These pieces of evidence support the protective role of ER signaling in attenuating HCC development or progression.

2.14. ERRs and ncRNAs

ERRs (ERRα, ERRβ, and ERRγ) are orphan NRs, sharing considerable structural homology to ERα and ERβ. ERRs are involved in tumor pathophysiology. Loss of ERRα, a master regulator in cellular energy metabolism, innate immunity, and cell differentiation, predisposes mice to carcinogens-induced development of HCC [200]. However, another study demonstrates that knock-down of ERRα inhibits cell proliferation, colony formation, invasion, and migration in HCC cells [201]. The IncRNA SPRY4-IT1 promotes ERRα expression, as the knock-down of SPRY4-IT1 suppresses the mRNA and protein expression of ERRα in HCC cells [201]. ERRγ is dysregulated by miRNAs in HCC cells. ERRγ is targeted by miR-940, a miRNA that is decreased in HCC tissues and cell lines, correlates with poor survival of HCC patients, and inhibits HCC cell growth through inducing apoptosis [202]. ERRγ is also targeted by miR-545 that can be a useful diagnostic marker for HCC [203]. ERRγ induces the expression of miR-433 and miR-127, which can be repressed by SHP [204].

2.15. 3-ketosteroid receptors and ncRNAs

3-ketosteroid receptors are steroid hormone receptors of the NR3 class. They act typically as dimeric entities, reside in the cytoplasm in the unliganded state, and are complexed with molecular chaperones. Upon agonist binding, they migrate to the nucleus to bind DNA elements and allow gene transcription to be regulated via interacting with other co-regulators of gene transcription, including RNA polymerase, acetyltransferases, and deacetylases [205]. 3-ketosteroid receptors include glucocorticoid receptor (GR, NR3C1), mineralocorticoid receptor (MR, NR3C2), progesterone receptor (PR, NR3C3), and AR (NR3C4).

GR imposes profound effects on various physiological processes, including metabolism, immunity, neuronal development, and adaptation to stress. MiR-124a is a GR-targeting miRNA that is up-regulated in betaine-exposed porcine neonatal livers and attenuates GR-mediated hepatic expression of lipogenic genes [206, 207]. Interestingly, the IncRNA GAS5 represses GR’s transcriptional activity via its decoy RNA “glucocorticoid response element”, thus playing a critical role in autoimmune, inflammatory, and infectious disease (Figure 2D–d) [39, 208]. GR can regulate the expression of a conserved miR-379/410 gene cluster. Hepatic miR-379 overexpression is sufficient to cause systemic dyslipidemia in mice via targeting hepatic lipolysis-stimulated lipoprotein receptor (LSR) and low-density lipoprotein receptor (LDLR) and impairing hepatic lipid re-uptake [209].

MR is a direct target of miR-766 and involved in miR-766-promoted proliferation and metastasis of HCC cells [210]. MR expression is significantly down-regulated in HCC tissues from patients. MR activates the expression of miR-338-3p that inhibits the Warburg effects by targeting the key glycolytic enzyme PKLR in HCC cells [211]. Therefore, it seems that MR plays a suppressive role in HCC progression.

In humans, PR is encoded by a single PGR gene which has two isoforms, PR-A and PR-B, to mediate the effects of progesterone oppositely. PR is likely a direct target of miR-26a but needs further validation [197].

AR signaling takes part in HCC initiation and progression, and its role appears to depend on the tumor stage. AR promotes HBV-, HCV-, and carcinogen-induced HCC initiation in genetically modified animals but suppresses HCC cell invasion and metastasis in vivo at a late stage [212]. Androgen increases the level of miR-216a to suppress tumor suppressor in lung cancer-1 (TSLC1) expression in HCC onset, which represents a new mechanism for the androgen pathway in early hepatocarcinogenesis in HBV-related male patients [213].

AR activation up-regulates levels of miR-22, miR-690, miR-122, let-7a, miR-30d, and let-7d and down-regulates the miR-22 target genes ERα and aromatase in female mouse livers, without affecting the levels of itself and ERβ, implying the crosstalk between sex hormone receptor signaling pathways [214]. Hypoxia decreases AR expression, which reduces the expression of miR-520f-3p and thus increases the expression of SOX9, a target of miR-520f-3p, in HepG2 cells. This AR/miR-520f-3p/SOX9 signaling might affect the chemosensitivity of HCC cells to sorafenib [212]. AR expression can be positively regulated by miR-367-3p which directly targets MDM2 3′-UTR to inhibit MDM2-mediated proteasomal degradation of AR [215].

2.16. Liver receptor homolog-1 (LRH1) and ncRNAs

LRH1 (NR5A2) is a critical regulator in embryonic development, cholesterol transport, bile acid homeostasis, and steroidogenesis. LRH1 regulates bile acid synthesis by regulating the expression of the rate-limiting enzyme CYP7A1. MiR-10a overexpression in hepatocytes blunts LRH1 expression and impairs bile acid synthesis [216]. LRH1 represses apoptosis, induces CYP19A1 (the key gene in E2 production) expression, and increases estrogen secretion in porcine granulosa cells; these can be antagonized by an LRH1-targeting miRNA, miR-1275 [217]. LRH1 is also targeted by miR-381 that suppresses Wnt/β-Catenin signaling to inhibit proliferation and invasion of HCC cells [218].

2.17. Dosage-sensitive sex reversal, adrenal hypoplasia critical region, on chromosome X, gene 1 (DAX-1) and ncRNAs

DAX-1 (NR0B1) is an atypical orphan NR that maintains the pluripotency of mouse embryonic stem cells (mESs). DAX-1 can form a complex with LRH1 and augments LRH1-mediated Oct4 activation in the IncRNA SRA-dependent manner in mES cells [219]. Interestingly, complexing with NCOA1, SRA serves as a transcriptional coactivator selectively for steroid hormone receptors, including AR, ERα, ERβ, PR, and GR [40]. DAX-1 also acts as a transcriptional corepressor. MiR-561 inhibits its target gene DAX-1 expression; this inhibition abolishes the interaction between DAX-1 and HNF4α, leads to HNF4α-mediated transactivation of PXR and CAR, and worsens APAP-induced hepatotoxicity [220].

2.18. SHP and ncRNAs

SHP (NR0B2) is a unique member of the NR superfamily, which interacts with a variety of other NRs including AR, ERα, HNF4α, LRH1, LXRα, PPARγ, RARα, RXRα. SHP contains dimerization and ligand-binding domains but lacks a DNA-binding domain. As an orphan NR, the role of SHP in transcriptional regulation, bile acid synthesis, lipid/glucose metabolism, drug metabolism, and liver pathophysiology has been extensively reviewed [221].

SHP is a transcriptional repressor and protects against cholestatic liver injury. Cholestasis decreases hepatic SHP protein levels and thus increases the expression of CYP2D6, a major drug-metabolizing enzyme, in cholic acid-fed mice, which is in part mediated by the upregulation of a SHP-targeting miRNA, miR-142-3p [222]. SHP inhibits the IncRNA MEG3 expression by repressing cAMP response element-binding protein-mediated transactivation of MEG3’s promoter. MEG3 is reactivated in fibrotic and cirrhotic livers. Forced overexpression of MEG3 in mouse livers causes rapid Shp mRNA decay, increases the expression Cyp7a1 and Cyp8b1, disrupts bile acid homeostasis, and induces cholestatic liver injury. Thus, SHP and MEG3 constitute a feedback loop of reciprocal inhibition to maintain bile acid homeostasis [223]. The IncRNA H19 is reactivated during cholestasis [12, 224, 225]. Exosomal-H19 derived from cholangiocytes or serum can be transported into hepatocytes to suppress SHP expression, promoting cholestatic liver injury in mice [225].

In Shp^−/− mice, a microarray profiling shows 21 up-regulated miRNAs clustered on chromosome 12, including miR-433 and miR-127, two miRNAs regulated by ERRγ [204], yet their function in bile acid metabolism needs to be determined. Opposing to the role of PPARα and LRH-1, SHP inhibits miR-200c expression to increase migration in HCC cells, suggesting that SHP may play a suppressive role in tumor progression [226].

3. Conclusions and Perspectives

NRs are essential in liver physiology and pathophysiology. Dysregulation of NRs contributes to the pathogenesis of a broad range of liver disorders. NRs are the pharmaceutical targets of several drugs being developed to treat liver disease [227]. The efficacy and safety of Elafibranor (GFT505), a dual PPARα/δ agonist, have been evaluated in patients with NASH in a phase III clinical trial (NCT02704403) because its agonism efficiently prevents the progression of NASH to liver fibrosis in mice [228]; however, the trial has been terminated because the study did not meet the predefined primary surrogate efficacy endpoint. TERN-101, a non-bile acid FXR agonist developed by Terns Pharmaceuticals, is under a phase II clinical trial in non-cirrhotic NASH patients in the U.S.

Targeting ncRNAs is an attractive strategy for diagnostic biomarker discoveries and therapeutics in liver disease. For instance, MRX34 is a liposome-formulated miR-34 mimic-based drug and has undergone a phase I clinical study in patients with advanced solid tumors, including HCC [229]. United States Food and Drug Administration (FDA) has already approved several small RNA (sRNA) drugs for clinical medical intervention, which paves the way for the development of more RNA-guided drugs [230]. Multiple miRNA drug candidates have undergone phase I or phase II clinical trials, but none is in clinicaltrials.gov database for phase III trials [231, 232]. MiRNAs can target the “difficult to drug” genes. Indeed, RNA-based therapeutics displays many superiors, with the ability to target more extensive druggable targets from proteins, RNAs, and the genome [230]. Due to the immune-related serious adverse events, the multicenter phase I study of MRX34 has been terminated (NCT01829971), suggesting that we still confront challenges regarding the distinct chemistry and pharmacokinetics properties and immunogenicity of sRNA molecules after bio- or chemo-engineering and when administrated in vivo, which has been elegantly elaborated elsewhere [233].

A perspective on the current status of miRNA-based diagnostics and therapeutics is that many pharmaceutical and biotech companies are developing miRNA panels in disease stratification and diagnosis. For instance, ThyraMIR from Interpace Diagnostics includes the quantification of 10 miRNAs to classify thyroid and pancreatic cancer [232]. On the other hand, the therapeutic market, mainly driven by miRNA mimics and antagomiR products, is less but actively advanced. Current active companies working on miRNA-based therapeutics and their respective products are introduced elsewhere [232]. RG-101 (produced by Regulus Therapeutics), a novel GalNAc (N-acetyl-D-galactosamine)-conjugated inhibitor of miR-122, demonstrates significant potency in reducing the load of HCV, representing a flagship product of this class of future drugs [232].

Mutations of NRs should be considered when designing ncRNA-guided trials if the action mechanism relies on base-pairing. In humans, genetic variants or dominant-negative mutations of PPARγ are associated explicitly with NAFLD and its progression towards inflammation and fibrosis [234]. Nanoparticles have been used for the delivery of miRNAs in vivo. The efficiency and distribution of a lipid nanoparticle-mediated delivery of miR-7 have been evaluated in animals [235]. Combining the RNA or therapeutic particles with a specific molecule that can specifically bind to cells of interest and be taken through endocytosis is another useful delivery approach. The miRNA panel- or network-based manipulation of NRs would advance single NR ligand-based therapeutics for avoiding side effects from the combination of NRs’ chemical agonists/antagonists.

Due to the prevalence of high-throughput sequencing, sRNA families have been rapidly extended. Emerging studies have identified tRNA-derived sRNAs (tsRNAs, also known as tRNA-derived fragments, tRFs) and rRNA-derived sRNAs (rsRNAs), two new sRNA categories mediating diverse functions. TsRNAs regulate ribosome biogenesis, protein translation, retrotransposon mobility, cell differentiation, epigenetic inheritance, cancer metastasis, and neurological diseases; rsRNAs are implicated in metabolic disorders and inflammation [236, 237]. Taking advantage of small RNA-seq technology, a study reveals that tsRNAs are dramatically increased in plasma exosomes of patients with liver cancer, demonstrating that plasma exosomal tsRNAs could serve as novel diagnostic biomarkers [238]. Inhibition of a specific tsRNA, LeuCAG3′tsRNA, induces apoptosis in rapidly dividing cells and a patient-derived orthotopic HCC model in mice [239].

Interestingly, snoRNAs, a class of ncRNAs primarily guiding nucleotide modifications of various RNAs, mostly on rRNAs, show miRNA-like function [240–242]. Loss of SNORA24, an H/ACA snoRNA, cooperates with RAS^G12V to promote the development of liver cancer in mouse models [243]. Overexpressing the snoRNA Jouvence which has not been annotated in the genome, leads to cell dedifferentiation and upregulates NR2F1 (chicken ovalbumin upstream promoter-transcription factor I, COUP-TFI) expression in human colon cancer HCT116 cells [244]. The regulatory role of snoRNAs in human diseases such as HCC has been elucidated before [245, 246]. The functional association between snoRNAs and NRs in liver pathophysiology has not been reported so far. We are still far away from the complete understanding of the diverse biological function and disease association of ncRNAs. Future research using omics tools will help identify critical ncRNA-NR circuits to develop biomarkers for the diagnosis and treatment of liver diseases.

Highlights.

• NcRNAs are critical regulators of gene expression via distinct action mechanisms

• Nuclear receptors are dysregulated by miRNAs and IncRNA in liver pathologies

• Nuclear receptors and ncRNAs regulate each other and form feedback

• Targeting ncRNAs and nuclear receptors is critical to the development of biomarkers

• Multiple ncRNA drug candidates have undergone clinical trials