Epigenetics of alcohol-related liver diseases

Summary

Alcohol-related liver disease (ARLD) is a primary cause of chronic liver disease in the United States. Despite advances in the diagnosis and management of ARLD, it remains a major public health problem associated with significant morbidity and mortality, emphasising the need to adopt novel approaches to the study of ARLD and its complications. Epigenetic changes are increasingly being recognised as contributing to the pathogenesis of multiple disease states. Harnessing the power of innovative technologies for the study of epigenetics (e.g., next-generation sequencing, DNA methylation assays, histone modification profiling and computational techniques like machine learning) has resulted in a seismic shift in our understanding of the pathophysiology of ARLD. Knowledge of these techniques and advances is of paramount importance for the practicing hepatologist and researchers alike. Accordingly, in this review article we will summarise the current knowledge about alcohol-induced epigenetic alterations in the context of ARLD, including but not limited to, DNA hyper/hypo methylation, histone modifications, changes in non-coding RNA, 3D chromatin architecture and enhancer-promoter interactions. Additionally, we will discuss the state-of-the-art techniques used in the study of ARLD (e.g. single-cell sequencing). We will also highlight the epigenetic regulation of chemokines and their proinflammatory role in the context of ARLD. Lastly, we will examine the clinical applications of epigenetics in the diagnosis and management of ARLD.

Key points.

• •The definition of epigenetics has evolved over time. Currently it refers to “the study of molecules and mechanisms that can perpetuate alternative gene activity states in the context of the same DNA sequence”
• •The epigenetic machinery controls gene expression by modulating chromatin confirmation, high order chromatin 3D structure, and the interaction between genes and the transcriptional apparatus.
• •Epigenetic dysregulation is central to the pathophysiology of ARLD.
• •Advances in the study of epigenetics and single-cell epigenome technique and their implications in health and disease have paved the way for novel diagnostic modalities (liquid biopsy) and therapeutic options (epidrugs).
• •Selective bromodomain inhibitors are a novel class of molecules with therapeutic potential in a wide range of liver diseases.

Introduction

Alcohol consumption is a leading cause of preventable death and is responsible for roughly 3.3 million deaths annually (5.9% of all deaths).¹ Alcohol-related liver disease (ARLD) is one of the most prevalent forms of liver disease in the world. ARLD represents a spectrum of disorders that encompasses alcohol-related fatty liver, alcohol-related hepatitis (AH)/steatohepatitis (ASH), alcohol-related cirrhosis and hepatocellular carcinoma (HCC). The natural history and pathophysiology of ARLD are complicated. The vast majority of chronic alcohol users develop alcohol-related fatty liver, but only a minority progress to alcohol-related cirrhosis or HCC.² Genetic and epigenetic factors, at least in part, determine disease onset and progression. For example, genome-wide association studies revealed multiple genes that were linked to the risk and severity of ARLD (e.g. PNPLA3, TM6SF2, MBOAT7, HSD17B13).[3], [4], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14]

The management of ARLD is determined by the extent of the disease. Abstinence, nutritional support, and screening for associated complications (e.g., HCC) represent the foundation of ARLD management.^2,[15], [16], [17] Agents like the tumour necrosis factor (TNF)-α inhibitors infliximab and etanercept have been used to treat AH based on their anti-inflammatory properties, but results have been disappointing.^18,19 Currently no agents are available that truly alter the outcome of advanced ARLD. Accordingly, liver transplantation is the only long-term management solution. Notably, ARLD accounted for 28% of all patients on the liver transplant waiting list in the US between 2006 and 2014.²⁰

The scarcity of available organs, the risk of relapse following transplantation, and the ‘self-inflicted’ and ‘moral failing’ view of ARLD raise numerous ethical questions, with the main question being ‘Is it fair to give patients with ARLD such a limited resource?’. Studies have shown that the majority of patients transplanted for ARLD have good outcomes with relatively low rates of relapse when proper selection criteria are applied (e.g. abstinence for >6 months, presence of appropriate social support…etc).^21,22

Given the significant burden associated with ARLD and limited treatment options, viewing ARLD through the lens of epigenetics is of paramount importance, particularly in the era of individualised and precision medicine. Understanding the intricate mechanisms that orchestrate the maintenance and reprogramming of the genetic code of the constituent cells of the liver in health and disease provides an unrivalled opportunity to prevent, better diagnose (e.g. liquid biopsies),[23], [24], [25], [26] and potentially reverse the deleterious effects of alcohol (e.g. epidrugs).^27,28

Herein, we will cover the basic epigenetic mechanisms while highlighting relevant examples from the realm of ARLD when applicable. Additionally, we will discuss the chromatic structure and enhancer-promoter (E-P) interactions and their role in ARLD. We will conclude by summarising the clinical applications of epigenetics in the field of ARLD.

Epigenetics: the writers, the readers, and the erasers

The definition of epigenetics has evolved over time, in keeping with our deepening understanding of cell fate, pluripotency, and plasticity. Originally, the term epigenetics referred to the process by which the genotype brings the phenotype into being. Currently, it refers to ‘the study of molecules and mechanisms that can perpetuate alternative gene activity states in the context of the same DNA sequence.’²⁹ The major epigenetic mechanisms include DNA methylation, post-translational modification of histones (methylation and acetylation of lysine and arginine, phosphorylation of serine and threonine, ubiquitination, and SUMOylation of lysine), ADP-ribosylation, histone replacement, and non-coding RNAs. Traditionally, epigenetic modifiers are classified into 3 groups: the writers, the readers, and the erasers. An in-depth description of the epigenetic machinery is outside the scope of this article. We refer the reader to the following recent publications for more details.[29], [30], [31], [32], [33] We also encourage the reader to refer to Fig. 1.

Fig. 1.

The epigenetic modifiers: writers, readers, and erasers.

DNA methylation is mediated by DNMT (writer). Methyl groups (M) are attached to CpG dinucleotide islands, reversed by TET (eraser). MECP2 (reader) recognises methyl groups and represses expression of associated genes. Histone methylation is mediated by HMT (writer). KDM and PADI4 (erasers) remove methyl groups from lysine and arginine, respectively. Tudor/PHD/chromodomain (readers) recognise methylated histones. Histone acetylation is mediated by HAT (writer) and reversed by HDAC (eraser). Acetyl groups are recognised by bromodomains (readers). Non-coding RNAs interact with the different epigenic modifiers and modulate their function. circRNA, circular RNA; DNMT, DNA methyltransferase; HAT, histone acetyltransferase; HDAC, histone deacetylase; HMT, histone methyltransferase; KDM, lysine demethylase; lncRNA, long non-coding RNA; MECP2, methyl-CpG binding protein 2; miRNA, microRNA; ncRNA, non-coding RNA; PADI4, peptidyl arginine deiminase 4; PHD, plant homeodomain; TET, ten eleven translocation.

DNA methylation, histone modifications and non-coding RNA in ARLD

DNA methylation

DNA methylation is one of the better-described epigenetic modifications. Methylation occurs mostly at the 5th carbon of cytosine within CpG dinucleotide-rich islands that predominantly occupy the 5′ promoter region of genes. S-Adenosyl-L-methionine is the methyl group donor. The deposition, removal, and maintenance of methyl groups is a dynamic process that is mediated by a group of enzymes of 2 broad types: DNA methyltransferases (DNMTs) and DNA demethylases.³³ Methylated CpG islands act as binding sites for repressive proteins with a methyl-binding domain (e.g. methyl-CpG binding protein 2 [MECP2]).³⁴ DNA methylation can be altered in response to alcohol consumption via different mechanisms, namely, direct inhibition of DNMTs, distortion of the 1-carbon metabolism cycle via changes in the intracellular redox balance, or limited dietary intake of folic acid (methyl donor).³⁵ Differential methylation of specific CpG dinucleotides in patients with ARLD and non-alcohol-related fatty liver disease (NAFLD) has been well described.^36,37

Alcohol-related steatosis and steatohepatitis and DNA methylation

Increased levels of DNA methylation were shown to be present in patients with alcohol-related disease by Bonsch et al. more than a decade ago.³⁸ In the subsequent years, numerous studies have linked DNA methylation and the development of ASH.[39], [40], [41], [42], [43] More recently, a study uncovered a new axis consisting of FKBP5-YAP-TEAD1-CXCL1 connecting ethanol consumption and the development of steatohepatitis.⁴⁴ FK506-binding protein 5 (FKBP5) is a cochaperone protein that is involved in stress-related disorders.⁴⁵ FKBP5 expression was shown to be upregulated in patients with ARLD and in ethanol-fed mice. When compared with Fkbp5 knockout mice, wild-type mice had higher expression of FKBP5 after eating ethanol-containing chow. Ethanol feeding also led to neutrophilic infiltration of the liver, which is a histologic hallmark of AH.[46], [47], [48] Intriguingly, this effect was attenuated in Fkbp5 knockout mice. In patients with ARLD, the promoter of FKBP5 was hypomethylated. By interacting with Yes-associated protein (YAP) and TEA domain transcription factor 1 (TEAD1), FKBP5 increases the expression of the inflammatory C-X-C motif chemokine ligand (CXCL)1. This axis consisting of FKBP5-YAP-TEAD1-CXCL1 may provide new targets for future treatments.⁴⁴

Liver fibrosis, ARLD and DNA methylation

Fibrosis is a common end pathway of many liver disease processes including ARLD.^43,[49], [50], [51] In a healthy liver, hepatic stellate cells (HSCs) are quiescent perisinusoidal cells, but in response to hepatic damage, HSCs undergo transdifferentiation into extracellular matrix-depositing myofibroblasts.^50,51 HSC transdifferentiation is mediated in part by DNA methylation.^50,52

In support of this finding, treatment of HSCs with the DNMT inhibitor azacitidine prevented HSC transdifferentiation in murine models.⁵³ Peroxisome proliferator activated receptor-γ (PPARG) is one of many genes that is differentially methylated during the process of HSC transdifferentiation. MECP2 promotes the methylation and repression of PPARG and inhibition of MECP2 was associated with reduced fibrosis in murine models.^54,55 Likewise, the histone methyltransferase G9a activity, alongside DNMT1, was linked to the fibrogenic activation of HSCs in patients with chronic liver injury including those with ARLD.⁵⁶ Intriguingly, the use of the novel dual G9a/DNMT1 inhibitor CM272 reduced the burden of fibrosis in mouse models.⁵⁶

In patients with either alcohol or non-alcohol-related liver disease, the same differential methylation pattern at the PPARG promoter was detectable in the pool of cell-free DNA. Moreover, the levels correlated with the degree of hepatic fibrosis.⁵⁷ Interestingly, the negative effects of toxins, including alcohol, can influence subsequent generations, a phenomenon known as transgenerational epigenetic inheritance.^58,59 Against this background, the offspring of rats with carbon tetrachloride–induced hepatic fibrosis demonstrated upregulation of PPARγ as an adaptive protective mechanism. These effects are believed to be mediated through DNA methylation and histone acetylation in the paternal sperm.⁶⁰

Post-translational histone modification

Histone methylation and acetylation have been studied extensively as markers of chromatin expression states.[61], [62], [63] Post-translational modification of histones affects DNA expression via 2 pathways: i) by changing the charge of histone proteins via acetylation, thus loosening the binding to nucleosomal DNA and leading to increased expression; and ii) by post-translational modifications acting as homing signals for proteins that can alter DNA expression.

Histone acetylation

Histone acetylation and deacetylation are catalysed by the enzymes histone acetyltransferase and histone deacetylase (HDAC), respectively.^32,64 Unlike histone methylation, histone acetylation is linked to transcriptional activation by promoting a less-taut 3D configuration of DNA or by acting as a signal for reader proteins (e.g., bromodomain [BRD]-containing proteins).³²

Ethanol is known to affect hepatocyte nuclear histone acetylation status in a time-and concentration-dependent manner.[65], [66], [67] For example, the expression of class I alcohol dehydrogenase (ADH1) – a key enzyme in the metabolism of alcohol – was upregulated in response to alcohol-containing chow in rats. This upregulation was associated with increased histone acetylation of the promoter region and coding region of the ADH1 gene.^66,68 Excessive alcohol use can lead to hepatic steatosis. Interestingly, binge alcohol treatment affected metabolic pathways controlling lipogenesis and fatty acid β-oxidation by deregulation of various HDACs.⁶⁹ Similarly, increased histone acetylation of the promoter region of PNPLA3 (patatin like phospholipase domain containing 3) was demonstrated in response to alcohol treatment in mouse models.⁷⁰ Sterol regulatory element-binding proteins (SREBPs) play a central role in cholesterol and lipid metabolism and dysregulation of SREBPs is associated with hepatic steatosis.⁷¹ Interestingly, SREBP-1 activity is augmented in response to alcohol treatment via increased histone acetylation.⁷² This effect was abolished following treatment with resveratrol, a potent sirtuin (SIRT)1 agonist.⁷² Along the same line, overexpression of SIRT2 mediated deacetylation of CCAAT/enhancer binding protein-β, which prevented alcohol-induced liver injury.⁷³ Likewise, repression of carnitine palmitoyltransferase-1 gene expression via the action of HDAC1 explains the mechanism that underpins the ethanol-mediated decrease in carnitine palmitoyltransferase-1 expression and alcohol-related steatosis. This effect was ameliorated following treatment with the HDAC1 inhibitor tributyrin.⁷⁴

Alcohol promotes inflammation.^75,76 In ARLD, alcohol erodes gut endothelial integrity, which leads to increased translocation of lipopolysaccharide (LPS) into the portal circulation.⁴⁸ Ethanol and its end product acetate directly affect macrophages’ response to LPS.^77,78 Macrophages cultured in methanol-containing medium, exhibited enhanced expression of interleukin (IL)-6, IL-8, and TNF-α after LPS stimulation.^77,78 Promoter sites of proinflammatory genes in alcohol-treated macrophages exhibited increased acetylation which can be attributed to reduced HDAC activity. This effect was prevented by inhibiting the metabolism of ethanol into acetate.⁷⁷ Curiously, the use of the SIRT1 inhibitor sirtinol augmented TNF-α release from LPS-treated macrophages.⁷⁸

Histone methylation

Histone methylation and demethylation at lysine and arginine residues of histones H3 and H4 are mediated by histone methyltransferase and lysine demethylase, respectively. Lysine and arginine can be monomethylated or dimethylated, and lysine can be trimethylated. The influence of lysine methylation on DNA expression is complex and depends on the lysine residue methylated.⁷⁹ Generally, methylation events occurring at some locations (e.g., H3K4, H3K36, and H3K79) lead to transcriptional activation, whereas methylation at H3K9, H3K27, and H4K20 is linked to transcriptional repression.⁸⁰ In the context of ARLD, multiple studies showed altered histone methylation in rat hepatocytes after ethanol treatment, with increased H3K4 dimethylation and decreased H3K9 dimethylation in one study.⁸¹ Notably, the changes observed in the histone methylation status were dependent on the mode of alcohol exposure, namely, acute binge model vs. chronic model.⁸² Other studies investigated the role of histone methylation in AH and fibrosis. For example, LPS leads to increased methylation of the TNF promoter region. This effect was abrogated by S-adenosyl L-methionine treatment.⁸³ Regarding fibrosis, a direct effect of ethanol was seen on HSCs. When cultured in ethanol-containing media, primary rat HSCs demonstrated increased expression of extracellular matrix-associated genes, including type I/III collagen, elastin, and tissue inhibitor of metalloproteinases.³⁷ MLL1 (KMT2A) and H3k4 methylation were enriched at the elastin gene in alcohol-treated HSCs.³⁷

Non-coding RNA

Non-coding RNA refers to a wide range of RNA molecules of varying lengths and functions (Fig. 2). We will focus on the role of microRNA (miRNA) and long non-coding RNA (lncRNA) in the pathogenesis of ARLD.

Fig. 2.

Role of ncRNA in alcohol-related liver disease.

NcRNAs are a group of RNA molecules of various length that are not translated into protein. The role of miRNA in the pathogenesis of alcohol-related liver disease has been reviewed extensively over the years.^14,35,[157], [158], [159], [160], [161] Dysregulation of miRNAs in alcohol-related liver disease is linked to the development of steatosis (yellow circle), inflammation via recruitment of inflammatory cells (blue circle), fibrosis (blue lines) and HCC (green circle). The associated table shows examples of dysregulated miRNAs in alcohol-related liver disease and their role in its pathogenesis. References (miR-21,[162], [163], [164] miR-26a,¹⁶⁵ miR-27a,^166,167 miR-29,¹⁶⁸ miR-34a,[169], [170], [171] miR-122,^95,172,173 mir155,^93,[174], [175], [176], [177] miR-125b,¹⁷⁸ miR-182,^179,180 miR-126,¹⁸¹ miR-200a,¹⁸² miR-181b-3p,¹⁸³ miR-214,¹⁸⁴ miR-199,¹⁸⁵ miR-217,⁹⁴ miR-223,¹⁸⁶ miR-291b,¹⁸⁷ miR-378¹⁸⁸). EMT, epithelial-mesenchymal transition; HCC, hepatocellular carcinoma; ncRNA, non-coding RNA.

MicroRNA and ARLD

Alcohol promotes inflammation, steatosis, and subsequently fibrosis by regulating multiple miRNAs.[84], [85], [86], [87], [88], [89], [90], [91] In alcohol-fed murine models, miR-132 and miR-155 levels were upregulated in Kupffer cells.⁹² MiR-155 enhances the proinflammatory effects of alcohol on Kupffer cells.⁹³ Overexpression of miR-217 worsens ASH through SIRT1 inhibition. Targeting miR- 217 using miRIDIAN hairpin inhibitor ameliorated these effects.⁹⁴ Multiple miRNAs are downregulated in the pathogenesis of ARLD (e.g., miR-122, miR-148a, miR-708).[95], [96], [97] miR-122 plays a protective role against alcohol-mediated liver injury by reducing the level of hypoxia inducible factor-1α (HIF1α). Little is known about the regulation of miR-122 in ARLD.⁹⁵ A recent study recognised the transcriptional regulator (GRHL2) to be responsible for the downregulation of miR-122 and subsequent increase in HIF1α in murine alcohol disease models.⁹⁵ Similarly, miR-148a protects against inflammasome activation and pyroptosis via thioredoxin-interacting protein inhibition. Forkhead box O1 and hepatocyte nuclear factor 4α (HNF4α) have recently been described as novel transcriptional regulators of miR-148a in the context of ARLD.^96,98 miR-148a has also been shown to regulate the expression of multiple enzymes essential for the metabolism of various substances including alcohol (e.g., cytochrome P450 and alcohol dehydrogenase 4.^98,99 Also, miR-708 is suggested to inhibit hepatic inflammation and steatosis through its effect on ZEB1.⁹⁷

lncRNA and ARLD

Not much is known about the contribution of lncRNA in the context of ARLD. Multiple studies have linked lncRNA to the development of ARLD and progression to HCC. Dou et al.¹⁰⁰ analysed the effect of alcohol on lncRNA expression profiles in a murine ARLD model. In total, 29 lncRNAs were identified, 17 of which were downregulated. Pathway analysis of the top 5 downregulated lncRNAs (mou_lnc_0610005C13Rik, mou_lnc_1700023H06Rik, mou _ lnc _ Gm12265, mou _ lnc _ AW495222(39,807), and mou_lnc_Gm45724) showed an association with alcohol-induced hepatic oxidative damage and cellular inflammation. Furthermore, 5 regulatory networks were constructed to provide a deeper understanding of the mechanism of action of these lncRNAs in ARLD, but validation studies are awaited.¹⁰⁰ Also, through its interaction with SIRT1, the lncRNA MALAT1 (metastasis-associated lung adenocarcinoma transcript 1) has been shown to propagate fibrosis and inflammation. During liver injury, lnc-MALAT1 is overexpressed. Lnc-MALAT1 binds SIRT1, leading to its inactivation, which subsequently activates HSCs and results in extracellular matrix deposition and fibrosis.¹⁰¹

Advances in the study of epigenetics in ARLD

Recent advances in the technologies utilised in the study of epigenetics improved our appreciation of the epigenetic landscape during health and disease. In the next section we will describe current (chromatin confirmation capture assays) and more advanced (single-cell epigenome assays) technologies used in the study of ARLD.

3D genome structure, E-P interactions, and epigenetic gene regulation in ARLD

Technologies that study the interactions between genetic loci in 3D chromatin configuration reshaped our thinking of the role epigenetics plays in certain disease states (Fig. 3).^33,[102], [103], [104] Chromatin state annotation was developed by analysing chromatin modification patterns and has been a powerful tool in the discovery of regulatory patterns.¹⁰⁵ ChromHMM – a java-based programme that uses multivariate Hidden Markov Mode – can recognise abnormal chromatin states and their correlations to biological functions from a large scale functional database, and can enable visualisation of the whole genome.¹⁰⁵ Fig. 4 is an example of the application of ChromHMM.¹⁰⁶

Fig. 3.

Chromosome conformation capture (3C) methods study the interactions between genetic loci in 3D structure.

3C methods study the interactions between genetic loci in 3D structure. These 3D interactions can be hundreds or thousands of bases to megabases apart in sequence and can heavily regulate gene expression. Most 3C assays share the same basic principles, mainly the fixation of genomic DNA-DNA and DNA-protein interactions through chemical crosslinking. The DNA is then digested, proteins disassociated, and the crosslinked fragments are ligated. Fragments are sequenced and mapped onto the genome to localise these interactions,¹⁰⁴ 3C, chromosome conformation capture; 4C, chromosome conformation capture-on-chip; 5C, carbon copy chromosome conformation capture; ChIA-PET, chromatin interaction analysis by paired-end tag sequencing; CTCF, CCCTC-binding factor; Hi-C, chromosome capture followed by high-throughput sequencing.

Fig. 4.

ChromHMM model and chromatin transcriptional states.

“ChromHMM model based on melanoma tumour samples. Emission profile from a 15-State LearnModel based on the 6 histone modifications studied. ChromHMM identifies functionally distinct chromatin states representing both repressive and active domains, such as polycomb repression (State 1), heterochromatic repression (State 5), active transcription (State 8 and 9) and active enhancers (State 13 and 14).” (From Terranova et al.¹⁰⁶).

Chromosome conformation capture assays and E-P interactions

Original chromosome conformation capture (3C) studies established the existence of E-P contacts.¹⁰⁴ Chromosome capture followed by high-throughput sequencing (Hi-C) data showed the enhancer and associated promoter interaction within the boundaries of a tissue topologically associating domain (TAD).¹⁰⁷ DNA sequences within a TAD can physically interact with each other more frequently than with sequences outside the TAD. TNF-α and LPS are the 2 main upstream regulators in the course of ARLD. There are a few studies on the effects of TNF-α and LPS on E-P interactions. In fact, our study showed that pre-existing loops within a TAD can affect TNF-α-dependent transcriptional regulation in ARLD for the first time.^120,154 Along the same lines, the human genes C-C motif chemokine ligand 2 (CCL2) and CXCLs responded to TNF-α signalling in ARLD and in Hi-C experiments designed to study dynamic chromatin interactions in primary human fibroblasts (IMR-90); the chemokine genes were arranged collinearly within the CCL2 or CXCL gene clusters, which reflected their relative spatial-temporal expression patterns.[108], [109], [110] The genes appear to rely on long-range enhancer and promoter DNA contacts. Unexpectedly, 3C, chromosome conformation capture-on-chip (4C), and Hi-C studies showed that TNF-α-responsive enhancers are prelooped with their target promoters before signalling. Such pre-existing chromatin looping, which also exists in other cell types with different extracellular signalling, is a strong indicator of gene induction (Fig. 5). These observations suggest that the 3D chromatin landscape is stable and can influence the selection or activation of target genes by a ubiquitous transcriptional activator in a cell-specific manner, with the spatiotemporal deposition of active histone modifications.[108], [109], [110], [111] The systematic mapping of chromatin loops by high-resolution Hi-C helps us to understand loop formation dynamics. Studies show that about a billion Hi-C ligation junctions are found per cell type, and up to 10,000 long-range contacts or loops were called per cell line.¹¹² Approximately 30% of the loops involved genes. Applying modified chromatin interaction analysis with a paired-end tag sequencing protocol and Hi-C also demonstrated that regulatory chromatin loops involve CCCTC-binding factor (CTCF). Cohesin, a chromosome-associated multi-subunit protein complex is critical and highly associated with looped enhancers (Fig. 5).¹¹³

Fig. 5.

Chromatin interactions around CXCL super-enhancer and co-induced gene promoter clusters.

(A) TNF-α–induced SE and its interaction with the promoters of gene clusters, including IL-8, CXCL1, CXCL2, and CXCL3. CTCF, NF-κB p65, and BRD4 peaks are observed at the super-enhancer and the promoters of all these genes. (B) A snapshot of ENCODE data on the same gene loci. Chromatin immunoprecipitation-sequencing of CTCF in HUVEC shows the proximity of CTCF binding sites to the transcription start sites of all CXCL genes. BRD, bromodomain-containing; ChIP-Seq, chromatin immunoprecipitation-sequencing; CTCF, CCCTC-binding factor; CXCL, C-X-C motif chemokine ligand; HUVEC, human umbilical vein endothelial cell; IL8, interleukin-8; SE, super-enhancer.

3D epigenomics and ARLD

Research into 3D gene regulation has greatly improved in the past few decades.¹¹⁴ One of the earliest topological analyses was a 3C study demonstrating that a variant destabilised an E-P loop with the OCA2 gene and caused its downregulation.¹¹⁵ Recently, researchers have also established defined physiologic responses that lead to dynamic activation of pre-existing E-P interactions and the formation of new E-P loops in liver samples in response to physiologic stimuli (e.g., diet).¹¹⁶ For example, response to a fat rich diet was mediated largely by activation of preformed E-P loops interacting with nuclear receptors including HNF4α.¹¹⁶ Peculiarly, studies on the 3D epigenome and transcriptome in AH are scarce. However, regulation of HNF4α E-P interactions through looping might be of relevance in ARLD. Analysis of RNA-sequencing of hepatic samples from patients with AH linked the development of AH to dysfunction of liver-enriched transcription factors with HNF4α being one of the most dysregulated.¹¹⁷ Two promoter-driven, HNF4α-spliced isoforms in hepatocytes have been studied in detail using multiple epigenetic approaches such as whole-genome DNA methylome analysis, chromatin immunoprecipitation-sequencing (ChIP-seq) of histone markers, and single-nucleotide variation analysis. These studies show that AH livers underwent major alterations in DNA methylation patterns that resulted in chromatin remodelling.¹¹⁷ For instance, HNF4α has 12 isoforms, which are expressed under the control of 2 promoters and result from alternative splicing. These isoforms can be categorised into 2 types; the adult isoforms, HNF4α-P1, and the foetal isoforms, HNF4α-P2, which are driven by a ∼45-kb upstream alternative promoter. The relevance of the P2 isoforms in adult human liver disease is not clear. The authors¹¹⁷ found that HNF4α-P1 mRNA was unchanged in AH, but expression of the HNF4α-P2 isoforms was significantly increased in livers from patients with AH. They¹¹⁷ showed that the expression of the lncRNA HNF4A-AS1, which uses the same P1 promoter region of HNF4α, was decreased in patients with AH. The function of this antisense lncRNA was not previously known and seemed to be related to HNF4α regulation and cell differentiation and possibly HNF4α E-P looping.¹¹⁸ Thus, targeting epigenetic drivers that modulate HNF4α-dependent gene expression could be beneficial in patients with AH.¹¹⁷

Another facet of epigenomic regulation in ARLD is the role of super-enhancers. Super-enhancers is a term that denotes ‘groups of putative enhancers in close genomic proximity with unusually high levels of Mediator binding, as measured by ChIP-seq’.¹¹⁹ Our group and others demonstrated activation of cytokine pathways and chemokine production in AH.^120,121 Our initial transcriptomic study showed remarkable changes in the transcriptome and epigenome of AH cirrhotic livers, which were also accompanied by the upregulation of several CXCL chemokines. By using 3C, 4C, and analysis of histone markers such as H3K27ac, H3K4m1, H3K4m3, along with NF-κB ChIP-seq, our group also identified the existence of a super-enhancer governing CXCL chemokines that is located upstream of the CXCL locus in liver cells (Fig. 5). Similarly, we identified H3K27ac enrichment on the promoter and super-enhancer of CXCL chemokines in response to TNF-α stimulation in AH livers. Interestingly, pharmacologic inhibition of NF-κB and bromodomain-containing (BRD)4 binding can attenuate TNF-α-induced H3K27ac enrichment and downregulate CXCL expression. These findings and the favourable effects of suppressing the CXCL super-enhancer highlight the significance of epigenetic regulation in AH and a potential new treatment approach.¹²⁰

Single-cell epigenome applications in ARLD

Traditionally, studies describing epigenetic changes and regulators in the pathogenesis of human disease have been limited to bulk assessment of tissues. Recently, one study combined single-cell RNA-sequencing data from healthy livers and peripheral immune cells to measure cell proportions in early AH, severe AH, HCV, HCV with cirrhosis, and NAFLD;¹²² these analyses showed that patients with severe AH had the greatest change in cell composition. In addition, this study also identified a new group of inflammatory macrophages that is increased in patients with HCV. Network and signalling analysis also found that these changes are highly correlated with liver function tests. This evidence proved that only single-cell RNA-sequencing technology can provide this kind of statistical power in clinical disease studies.¹²² Although other techniques, such as assay for transposase-accessible chromatin (ATAC)-sequencing, can provide more useful information about the overall chromatic accessibility state, they are limited by tissue and disease heterogeneity.¹²³ Liver tissue heterogeneity is being increasingly appreciated thanks to new state-of-the-art technologies that allow disease conditions to be discerned at the single-cell level. Similarly, the pathologic process of ARLD includes injury not only to hepatocytes but also to HSCs, Kupffer cells, liver sinusoidal endothelial cells, and others.^120,124,125 Until recently, few technologies were available that enabled the determination of epigenomic changes in individual cell types. Single-cell technologies are now available such as RNA-sequencing for the transcriptome, ATAC-sequencing for chromatin-accessibility and potential epigenomic regulatory elements, and single-nucleus multiome that combines RNA and ATAC-sequencing data from individual nuclei. These technologies may also help explain differences in presentation and outcome in patients at different points on the ARLD spectrum.[126], [127], [128]

Response to injury and fate of various cell lineages are determined in part by sequence-specific transcription factors interacting with cis-regulatory elements in a cell- and tissue-dependent manner. This is a guiding principle to understanding heterogeneity in normal and diseased tissue. Single-cell ATAC and DNase (DNase I–hypersensitive sites) sequencing leverage the hypersensitivity of cis-regulatory elements to transposases and nucleases in poised-to-act or active states and can be used to generate genome-wide regulome maps.¹²⁹ Some other nuanced and less-widely used technologies such as single-cell transposome hypersensitive site sequencing, or studying individual cells using cells isolated via microfluidic devices or nanowell arrays, are also now available for the study of chromatin landscapes.[130], [131], [132]

As noted earlier, dysregulation of master transcription factors such as HNF4a is well described.¹¹⁷ Furthermore, an altered immune response in ARLD has been shown to have an epigenetic reprogramming function.^133,134 Transcription factors from the ETS, CCAAT/enhancer binding protein, and interferon-regulatory factor 1 families have been implicated in these changes.¹³⁴ Similarly, endothelial GATA4 has been shown to control liver fibrosis and regeneration by preventing a pathogenic switch in angiocrine signalling.¹³⁵ These cis-regulatory elements then lead to tissue-specific alterations in the transcriptome. Thus, identifying the role of one or a group of transcription factors in each cell type will help to identify new therapeutic targets, monitor responses, and provide insight into cell-cell interactions. Meanwhile, a lot of effort has also been put into the study of these pioneer transcription factors in cis-regulatory elements and E-P regulation, which are being recognised as druggable targets against disease onset and progression,^120,136,137 owing to their activity in a cell-identity and state-dependent manner. Advances in single-cell technologies will facilitate recognition of these interactions on a genome-wide level and provide new epigenomic target regions for known and novel genes of interest.

Clinical implications of the study of epigenetics in ARLD

The study of the epigenetics of ARLD has led to discoveries that are now entering everyday clinical practice in the form of either diagnostic tests or medications.

Epigenetics and the diagnosis of liver disease: Liquid biopsy

Liver biopsy remains the standard for the diagnosis and staging of acute and chronic liver diseases. The search for reliable non-invasive methods to diagnose and monitor disease progression has always been at the forefront of medical research.^25,88 Candidate serum biomarkers should fulfil certain criteria – they should be sensitive and specific and should correlate well with tissue-based tests. Liquid biopsies can be broadly defined as any body fluid-derived biomarker that can inform medical decision-making.¹³⁸ One example of liquid biopsy is miRNA and lncRNA profiling. In a recent study, Eguchi et al.¹³⁹ showed a specific miRNA signature that was released from hepatocytes during early ASH in a mouse model. Specifically, miRNAs Let7f, miR-29a, and miR-340 were increased in ASH mice but not in other chronic liver injury models. The same 3 miRNAs were increased in the serum of patients with mild ARLD.¹³⁹ Similarly, global profiling of sera from patients with and without ARLD showed a unique lncRNA signature. Further analysis identified 244 upregulated lncRNAs; lncRNAs AK128652 and AK054921 were significantly increased. To determine the prognostic value of AK128652 and AK054921, 48 patients with alcohol-related cirrhosis were followed up for 520 days, and these 2 lncRNAs were linked to shortened survival.¹⁴⁰

Detecting and staging the degree of hepatic fibrosis is essential for practicing hepatologists. Given the invasive nature of liver biopsy, many alternatives have been sought.¹⁴¹ As mentioned previously, PPARG is methylated during the transdifferentiation of HSCs into activated HSCs in the context of hepatic fibrosis. DNA methylation of the PPARG promoter was detected in cell-free DNA in patients with ARLD,¹⁴² and the level correlated with progression to cirrhosis in ARLD and NAFLD. Also, the hypermethylation of PPARG was specific to liver fibrosis. These findings are promising and may herald the development of cost-effective blood-based liquid biomarkers for the assessment of liver fibrosis.¹⁴²

Drugs targeting epigenetic regulation and ARLDs

Epigenetic drugs (epidrugs) are a group of compounds that target perturbed epigenetic changes in different disease states. The first class of epidrugs, DNMT inhibitors (e.g., azacitidine, decitabine), were in use for many years before their epigenetic mechanism of action was elucidated.²⁷ Intriguingly, the list of medications with previously unknown epigenetic modulatory function is expanding, which has broadened their therapeutic indications. A noteworthy example is the antiepileptic drug valproic acid, which has been shown to have HDAC inhibitor capabilities, with possible implications for the treatment of HCC¹⁴³ (Fig. 6). ARLD is a leading cause of HCC.¹⁴⁴ The role of environmental factors – including alcohol consumption – and the epigenetic changes that promote the development of HCC have been studied extensively.^42,144,145 The role of epidrugs in the management of liver disease is most established in the field of HCC. This topic has been reviewed recently by Fernandez-Barrena et al.²⁸ The field of epidrugs for the treatment of ARLD and NAFLD is not as developed.²⁸ However, recent advances have been made, particularly with the use of novel selective BRD inhibitors to counteract inflammation in ARLD. Chemokines are small chemotactic molecules that promote inflammation. In ARLD, CXCL chemokines facilitate neutrophil tissue infiltration and are linked to poor clinical outcome.

Fig. 6.

Epidrugs in alcohol-related liver disease.

Based on their mechanism of action, epidrugs can be divided into 8 broad categories: DNMTi, HATi, HDACi, HMTi, histone demethylase inhibitor, proteins binding to methylated histones inhibitor, proteins binding to acetylated histones inhibitor, and ncRNAs (such as antisense-RNAs, small interfering RNAs, and miRNAs). Ac, acetyl group; BETi, bromodomain and extraterminal motif inhibitor; circRNA, circular RNA; DNMT(i), DNA methyltransferase (inhibitor); HAT(i), histone acetyltransferase (inhibitor); HDAC(i), histone deacetylase; HMT(i), histone methyltransferase (inhibitor); KDM, lysine demethylase; lncRNA, long non-coding RNA; MECP2, methyl-CpG binding protein 2; miRNA, microRNA; ncRNA, non-coding RNA; PADI4, peptidyl arginine deiminase 4; PHD, plant homeodomain; PRMT, protein arginine methyltransferase; TET, ten eleven translocation.

Suppression of CXCLs mitigated alcohol-induced liver injury in mouse models.¹⁴⁶ The regulation and function of chemokines in liver disease was reviewed thoroughly by Cao et al.¹⁴⁶

As described previously, our group described the role of super-enhancers in regulating inflammation in ARLD, particularly upregulation of several CXCL chemokines.¹²⁰ The bromodomain and extraterminal (BET) family comprises epigenetic reader proteins. BET proteins are known to bind with super-enhancers and modulate super-enhancers’ function in inflammatory conditions.^111,[147], [148], [149], [150] BET proteins recognise the acetyl group on tagged histone lysine residues through their BRD. Four BET proteins have been described, one of which is germ cell-specific (BRDT), with the other 3 being expressed ubiquitously (BRD2, BRD3, and BRD4). Each BET protein contains 2 BRDs that are structurally homologous (BD1 and BD2). The role of BET proteins in modulating the expression of inflammatory mediators drew attention to their therapeutic potential, leading to the design of multiple BET inhibitors (BETis). The original BETi molecules targeted both BD1 and BD2 in a non-selective manner.^151,152 The lack of selectivity limited the development of medications owing to the pleiotropic effects exerted by these inhibitors.¹⁵³ Accordingly, the development of selective BD1 and BD2 BETis represents a turning point in our understanding of the roles of BET proteins in health and disease states. In fact, our group found suppression of CXCL expression in mice undergoing alcohol binges/LPS injection using the novel selective BD1 inhibitor, UMN627.¹²⁰ UMN627 is a novel BD1-selective inhibitor, with a 20-fold higher affinity for BRD4 BD1 over BRD4 BD2.^154,155 Treatment with UMN627 not only decreased the expression of CXCL but also attenuated neutrophil infiltration.¹²⁰ Another relevant example is the BRD2-selective BETi GSK620 developed by Gilan et al.¹⁵³ In NAFLD mouse models, GSK620 resulted in reduced levels of steatosis, lobular inflammation, and hepatocyte ballooning.¹⁵³ Taken together, the findings of our group¹²⁰ and Gilan et al.¹⁵³ may facilitate a new era of therapeutic approaches for the treatment of liver disease. In contrast, salvianolic acid A (SAA), a phenolic acid compound found in Danshen (used in Chinese herbal medicine), is a non-specific BRD4 inhibitor. In a recent study, SAA appeared to protect against AH and fibrosis in mouse models.¹⁵⁶ The effects of SAA were mediated by the inhibition of BRD4 and the translocation of its downstream inflammatory mediator, high-mobility group box protein 1 (HMGB1), which is secreted primarily by inflammatory cells. HMGB1 is believed to regulate NF-kβ via the modulation of Toll-like receptor (TLR)2, TLR4, and TLR9, which subsequently culminates in the expression of inflammatory mediators such as IL-6 and TNF-α.¹⁵⁶

Conclusion

In the past couple of decades, our understanding of epigenetics and its molecular mechanisms has increased significantly. This knowledge has added a new layer to our understanding of disease mechanisms, facilitated innovative diagnostic avenues and capabilities, and, most intriguingly, paved the way for a new class of medications (epidrugs). Despite all these advances, the field of epigenetics and its clinical applications are still in their early stages. For example, adverse reactions to epidrugs have emerged because they have pleiotropic effects and off-target issues. Furthermore, advances in single-cell epigenomics will allow for recognition of these interactions on a genome-wide level and provide new target regions for known and novel genes of interest. A great deal remains to be determined, but the advances made thus far are promising.

Financial support

This work is supported by funding provided by the National Institutes of Health (USA NIH): R01 AA21171 and R01 DK59615 (V.H.S.)

Authors’ contributions

N.W.S. contributed to literature analysis, interpretation, intellectual input, drafting and editing of the manuscript. T.S.S. contributed to drafting part of the manuscript. V.H.S. and S.C. contributed to writing conception and design, analysis and interpretation of literatures, intellectual input, and editing of the manuscript.

Conflict of interest

The authors declare no conflicts of interest that pertain to this work.

Please refer to the accompanying ICMJE disclosure forms for further details.

Supplementary data

The following are the supplementary data to this article: