Genetic predisposition similarities between NASH and ASH: Identification of new therapeutic targets.

Fatty liver disease can be triggered by a combination of excess alcohol, dysmetabolism and other environmental cues, which can lead to steatohepatitis and can evolve to acute/chronic liver failure and hepatocellular carcinoma, especially in the presence of shared inherited determinants. The recent identification of the genetic causes of steatohepatitis is revealing new avenues for more effective risk stratification. Discovery of the mechanisms underpinning the detrimental effect of causal mutations has led to some breakthroughs in the comprehension of the pathophysiology of steatohepatitis. Thanks to this approach, hepatocellular fat accumulation, altered lipid droplet remodelling and lipotoxicity have now taken centre stage, while the role of adiposity and gut-liver axis alterations have been independently validated. This process could ignite a virtuous research cycle that, starting from human genomics, through omics approaches, molecular genetics and disease models, may lead to the development of new therapeutics targeted to patients at higher risk. Herein, we also review how this knowledge has been applied to: a) the study of the main PNPLA3 I148M risk variant, up to the stage of the first in-human therapeutic trials; b) highlight a role of MBOAT7 downregulation and lysophosphatidyl-inositol in steatohepatitis; c) identify IL-32 as a candidate mediator linking lipotoxicity to inflammation and liver disease. Although this precision medicine drug discovery pipeline is mainly being applied to non-alcoholic steatohepatitis, there is hope that successful products could be repurposed to treat alcohol-related liver disease as well.

Alcohol-related and non-alcoholic steatohepatitis share the main genetic determinants.

New drugs targeting steatohepatitis can be developed by integrating genetic, transcriptomic, proteomic and lipidomic data with experimental models.

Genetic-based drug discovery can highlight targets to treat both non-alcoholic and alcoholic steatohepatitis.

Derangements in genes that regulate hepatic fat accumulation and hepatocellular lipid droplet remodelling are key in the pathogenesis of steatohepatitis.

The PNPLA3 I148M variant is the main common genetic determinant of steatohepatitis.

PNPLA3 silencing has a beneficial impact on experimental steatohepatitis.

Downregulation of MBOAT7 impairs the synthesis of arachidonic acid-containing phosphatidyl-inositol and determines the accumulation of lysophosphatidyl-inositol, leading to steatosis and promoting steatohepatitis.

IL-32 is induced by lipotoxicity in hepatocytes and acts as a mediator of inflammation and steatohepatitis, representing a candidate biomarker and therapeutic target.

Introduction

Non-alcoholic and alcohol-related fatty liver disease (namely NAFLD and AFLD,) are already undoubtedly, and will continue to be, leading drivers of progressive liver disease and hepatocellular carcinoma (HCC) worldwide.[3, [4], [5], [6] Severe alcohol abuse leads to accelerated disease progression with higher rates of HCC, liver-related deaths and a poorer prognosis., On the contrary, NAFLD is most frequently related to metabolic dysfunction (MAFLD: metabolic dysfunction-associated fatty liver disease), and is associated with increased risk of cardiometabolic disease and cancer. The exact impact of alcohol abuse on the risk of chronic liver disease in the general population is difficult to assess, however alcohol-related cirrhosis is responsible for one-third of liver transplantations in the USA and about 20% in Europe; on the other hand, NAFLD is estimated to affect 25% of the global population, and its prevalence is expected to increase., For these reasons, fatty liver disease (FLD) will likely pose a significant threat to public health in the near future. Excess fat accumulation in hepatocytes is the hallmark of these conditions, and it has recently been implicated in the development of liver injury and disease. Although the main environmental triggers of fat accumulation differ between alcoholic fatty liver disease (AFLD) and NAFLD, they are frequently superimposed, and the pathogenesis of inflammation and progressive liver damage share many mechanisms. Throughout the manuscript, we will therefore commonly refer to these disorders as FLD. The pathogenic background shared among FLDs results in a phenotype which reflects metabolic aberrations, usually characterised by short-circuits in the hepatic lipid metabolism that mediate the enhancement of lipid storage in the liver parenchyma., The metabolic switching converts the liver to a lipid storing organ, a choice that is imposed by the necessity to counteract the accumulation of cytoplasmic free fatty acids (FFAs). The latter is mainly caused by the continuous uptake of lipids originating from the inflamed adipose tissue and by enhanced de novo lipogenesis (DNL), which is activated by impaired cellular redox potential and/or hyperinsulinemia.,, In this scenario, triglyceride (or triacylglycerol TAG) storage may be seen as a protective mechanism to counteract lipotoxicity in hepatocytes. During alcohol abuse, hepatic lipid accumulation is promoted through four main mechanisms (summarised in Fig. 1): i) increased production of NADH secondary to the oxidation of ethanol, which promotes FFA and TAG synthesis and inhibits mitochondrial β-oxidation; ii) increased uptake of FFA coming from chylomicrons secreted by the intestinal mucosa; iii) ethanol-mediated inhibition of AMP-activated protein kinase, with the consequent increase of DNL and reduction of lipolysis due to the downregulation of the peroxisome proliferator-activated receptor a (PPAR-α) and the upregulation of the sterol regulatory element-binding protein 1c (SREBP-1c); iv) mitochondrial and microtubular damage induced by acetaldehyde and subsequent reduction of NADH oxidation, which determines accumulation of very low-density lipoproteins.[16], [17], [18]

Fig. 1

Ethanol metabolism and related mechanisms promoting hepatic lipids accumulation.

During alcohol consumption, ethanol is oxidised to acetaldehyde (by the constitutive pathway involving the NAD-dependent ADH) or metabolised at the level of the microsomal system through an inducible NADPH-dependent pathway involving cytochrome P450 (MEOS). Both lead to the formation of acetaldehyde, which is subsequently metabolised to acetic acid by the 2-ADH. ADH, alcohol dehydrogenase; 2-ADH, 2-alcohol dehydrogenase; AMPK, AMP-activated kinase; FFAs, free fatty acids; PPAR-α, peroxisome proliferator-activated receptor a; SREBP-1c, sterol regulatory element-binding protein 1c.

The progression of liver damage accelerates when, especially at times of acute insults during the natural history of the disease, excess fat and lipotoxicity lead to inflammation, hepatocellular damage and fibrogenesis, a condition referred to as “steatohepatitis” (i.e. non-alcoholic steatohepatitis NASH and acute alcohol-related steatohepatitis ASH respectively)., Unfortunately, classical biomarkers cannot accurately stratify the risk of progressive liver disease, and there are no approved drugs to treat FLD. Indeed, the first drugs reaching late stage development for NASH either failed or were burdened by an unfavourable side effect profile. Innovative approaches for biomarker discovery and therapeutic target identification are therefore urgently needed. While recent attempts mostly exploited previous knowledge related to biomarkers and candidate drugs from other hepatic and metabolic diseases, human genetics is unravelling – through unbiased approaches – the specific determinants of FLD and related biological pathways. At the same time, genetics is highlighting the sources of heterogeneity and possible ways to personalise treatment. Nowadays, molecular genetics is providing new tools to directly manipulate a wide array of potential disease pathways in the liver., In this review, we will discuss the common genetic mechanisms underlying both NASH and ASH, and how they can be used to guide the design of innovative therapeutic approaches.

AFLD and NAFLD share genetic predisposition

The predisposition to develop progressive FLD has a strong inherited component.,, Remarkably, NAFLD and AFLD, as well as other liver diseases where hepatic fat accumulation plays a major role (e.g. for chronic hepatitis C), share most of the main genetic determinants (shown in Fig. 2).

Fig. 2

Alcohol-related and non-alcoholic steatohepatitis share major genetic determinants.

Among the main genetic determinants which predispose to the development of progressive FLD, it is possible to distinguish a central core of genes that include modulators of fatty acid metabolism, lipid storage and secretion. These genes highlight the common genetic background, shared between AFLD and NAFLD. Conversely, other genes are specifically associated with only 1 type of liver injury such as GCKR and APOB in NAFLD/MAFLD. AFLD, alcoholic fatty liver disease; APOB, apolipoprotein B; FLD, fatty liver disease; GCKR, glucokinase regulator; HSD17B13, hydroxysteroid 17-beta dehydrogenase 13; MAFLD, metabolic dysfunction-associated fatty liver disease; MARC1, mitochondrial amidoxime reducing component 1; MBOAT7, membrane bound O-acyltransferase domain containing 7; NAFLD, non-alcoholic fatty liver disease; PCSK7, proprotein convertase subtilisin/kexin Type 7; PNPLA3, patatin like phospholipase domain containing 3; TM6SF2, transmembrane 6 superfamily member 2.

The main genetic determinant of FLD is the rs738409 C>G variant, encoding the I148M protein variant of patatin like phospholipase domain containing 3 (PNPLA3). This single-nucleotide polymorphism has been strongly associated with both NAFLD, and AFLD, including progression to cirrhosis,, HCC, and severe ASH., During insulin-resistance, PNPLA3 is induced by insulin in hepatocytes, hepatic stellate cells (HSCs) and adipocytes under the control of SREBP-1c, whose activation is also promoted by ethanol feeding in murine models. PNPLA3 protein localises on lipid droplets (LDs), it has lipase activity and is involved in the remodelling of phospholipids and TAGs., The I148M variant induces a loss of function, but the mutated protein evades ubiquitination and accumulates on LDs; by sequestering ABHD5/CGI-58, the variant also inhibits the activity of adipose tissue triglyceride lipase (ATGL/PNPLA2), the main LD-TAG-lipase in the liver. The result is the impairment of lipid turnover, the enlargement of LDs and lipotoxicity. In addition, the I148M variant may have independent effects on inflammation and fibrosis by impeding retinol release and rewiring transcriptional circuits in HSCs in response to liver damage.,,

A few other variants have been demonstrated to play a role in FLD. rs58542926 C>T encoding the TM6SF2 E167K variant causes hepatic fat accumulation, NASH, and fibrosis,, but is also associated with HCC development in AFLD. TM6SF2 is a Golgi membrane protein. The loss of function induced by the risk variant impairs very low-density lipoprotein secretion, with consequent TAG accumulation and heightened susceptibility to liver damage.,, Although upregulation of TM6SF2 may improve hepatic damage induced by FLD, it is not a viable therapeutic option, as it would increase circulating lipoproteins and cardiovascular disease.,, Glucokinase regulator (GCKR) variation is associated with FLD and fibrosis.[45], [46], [47] The GCKR protein acts as an inhibitor of glucokinase. The rs1260326 C>T variant encodes P446L, which lacks the ability to inhibit glucokinase in response to fructose-6-phosphate and to restrain glucose uptake, thereby favouring glycolysis, DNL and hepatic fat accumulation. Possibly because the impact of the variant on FLD depends on glucose levels, this was not reported to influence AFLD. Furthermore, GCKR cannot be targeted to reduce liver fat because this would lead to hyperglycaemia. On the other hand, genetic variation of membrane bound O-acyltransferase domain-containing 7 (MBOAT7) has been linked to predisposition to hepatic fat accumulation in both to AFLD and NAFLD.,, Hydroxysteroid 17-beta dehydrogenase 13 (HSD17B13) variants (rs143404524 and rs72613567) have more recently been implicated in the protection against ASH and NASH, cirrhosis and HCC.[52], [53], [54] The protection against liver damage was larger in PNPLA3 I148M variant carriers. HSD17B13 is involved in qualitative LD remodelling through conversion of retinol to retinoic acid in hepatocytes, but it has also been predicted to metabolise several lipid species. The protective variants have been shown to result in loss-of-function of enzymatic activity, suggesting that HSD17B13 inhibition may be considered as a therapeutic strategy. The rs2642438 mitochondrial amidoxime reducing component 1 (MARC1) variant (M187K) has also been identified as protective against FLD progression and cirrhosis.,

Once more, these data underline the common pathogenesis of FLD, highlighting possible common approaches for risk stratification and therapeutic intervention. In addition, the impact of variants in PNPLA3, TM6SF2, MBOAT7 and GCKR on hepatic inflammation and fibrosis was proportional to their effect on hepatic fat accumulation, suggesting that genetically determined hepatic fat levels predispose individuals to FLD and drive progression to fibrosis and HCC., Exploiting polygenic risk scores (PRS) as lifelong proxies of exposure to increased hepatic fat content and of qualitative alterations of hepatic fat, it was suggested that these variants have a causal role in determining progressive liver disease, and HCC. This “Mendelian randomization” framework exploits naturally occurring genetic variants, that have functional effects and are randomly inherited at conception, as experimental instruments to estimate the impact of the manipulation of the encoded proteins on human health. Since steatosis is the necessary condition for FLD onset and progression, hepatic fat is naturally the main therapeutic target (Table 1). Indeed, to date, pharmacological approaches targeting hepatic fat have generated the best results for the treatment of NASH.

Table 1

Examples of genetic pathways underlying fatty liver disease and correspondent potential therapeutic strategies.

Pathway	Genes	Therapeutic strategy	Development stage	Selected references
Hepatocellular fat accumulation	PNPLA3, TM6SF2, GCKR, GPAM, PYGO1, MBOAT7, APOB	Inhibition of lipogenesis (ACC, SCD1, DGAT1/2 inhibitors), stimulation of b-oxidation (TR-β agonists, PPAR agonists), anti-obesity drugs (targeting GLP1R), insulin sensitizers, glucose lowering drugs in T2D∗; any drug reducing hepatic fat	Preclinical, Phase I-IV	11, 47, 61, 62
Lipoproteins remodeling and cholesterol metabolism	PCSK7, PCKS9, APOE	PCSK9/7 silencing or neutralization, statins	PCKS9 approved for hypercholesterolemia	61, 69, 81, 135
Lipid droplet remodeling and lipotoxicity	PNPLA3, ABHD5;	PNPLA3 silencing	Preclinical, Phase I	11, 25, 47, 74, 76
	MBOAT7	Modulation of LPI metabolism, hepatic GPR55 antagonism	Hypothetical
		IL32 silencing or neutralization	Hypothetical	107
Oxidative and ER stress	HFE, MARC1, SOD2, UCP2, SERPINA1	Iron depletion, vitamin E, sylibinin	Phase II	55, 115, 136, 137
Hepatic retinol metabolism, inflammation and fibrogenesis	PNPLA3, HSD17B13, MERTK, LEPR	Retinoid receptors modulation, HSD17B13 silencing or direct inhibition, modulation of MERTK activity	Hypothetical	37, 39, [52], [53], [54], 62, 79, 80
Bile acids – FGF19 – b-Klotho – FXR axis	NR1H4, KLB	FXR agonists, FGF19 partial agonists, bile acids reuptake inhibitors, pre/pro-biotics, lubiprostone	Preclinical, Phase I-III	126, 129

ABHD5, abhydrolase domain containing 5; ACC, acetyl-CoA carboxylase; APOB, apolipoprotein B; DGAT, diacylglycerol O-acyltransferase; FGF19, fibroblast growth factor 19; FXR, farnesoid X receptor; GCKR, glucokinase regulator; GLP1R, glucagon-like peptide 1 receptor; GPR55, G protein-coupled receptor 55; HFE, homeostatic iron regulator; HSD17B13, hydroxysteroid 17-beta dehydrogenase 13; IL32, interleukin 32; KLB, klotho beta; LPI, lysophosphatidylinositol; MARC1, mitochondrial amidoxime reducing component 1; MBOAT7, membrane bound O-acyltransferase domain containing 7; NR1H4, nuclear receptor subfamily 1 group H member 4; PCSK, proprotein convertase subtilisin/kexin; PNPLA3, patatin like phospholipase domain containing 3; PPAR, peroxisome proliferator activated receptor; SCD1, stearoyl-CoA desaturase; SERPINA1, serpin family A member 1; SOD2, superoxide dismutase 2; T2D, type 2 diabetes; TM6SF2, transmembrane 6 superfamily member 2; TR-b, thyroid hormone receptor b; UCP2, uncoupling protein 2.

Stimulation of lipid secretion is not a viable option due to the major impact on circulating lipids and on cardiovascular risk profile in highly susceptible individuals.

The list of genetic FLD determinants is increasing by the day. Recently, an exome-wide association study of alanine aminotransferases in the population identified variants in apolipoprotein E (APOE) and glycerol-3-phosphate aminotransferase (GPAM) regulating hepatic lipid metabolism as risk factors for FLD and cirrhosis. Furthermore, in a case-control study, variants in the leptin receptor (LEPR), regulating appetite and liver fibrogenenesis, and in pygopus homolog-1 (PYGO1), potentially involved in LD remodelling, have been associated with clinical NAFLD. Several candidate genes involved in alcohol metabolism, ethanol-induced oxidative stress, inflammation and fibrosis have also been investigated as modifiers of AFLD risk, but additional data are needed to validate these findings.,

A new discovery paradigm: From human to molecular genetics and into the clinic

Directly targeting variant proteins responsible for the development of progressive steatohepatitis is an attractive approach for the treatment of FLD. Indeed, drug target discovery based on human genetics has a 4-fold higher probability of success (regulatory approval) than other methods. This new discovery cycle (Fig. 3) would start from human genetics, highlighting causal genetic variants robustly associated with the trait of interest. The subsequent linkage of genetic variation with transcriptomic, proteomic and – especially relevant here – lipidomic data,, coupled with basic studies examining the direct impact of variants in experimental models in vitro and in vivo, would highlight the mechanisms underlying the association. Genome editing techniques are proving to be powerful tools to model the impact of genetic variation in in vitro and in vivo models.[68], [69], [70] This may enable the selection of the most appropriate, if any, therapeutic target, to be tested first in models and, if successful, in patients (Fig. 3). In parallel, this process would lead to a progressive improvement in risk stratification, and in a more accurate selection of those who are more likely to benefit from the eventual therapeutic innovations (Fig. 3). Recent advances in genotyping and sequencing technologies, enabling the genome-wide characterisation of large cohorts of well-phenotyped individuals, alongside the computational power to link the different “omics” approaches are rendering this revolution possible.

Fig. 3

The new discovery paradigm: From human to molecular genetics and into the clinic.

FLD risk stratification results from a cyclic interplay between clinical studies in at risk individuals and the manipulation of the associated pathways to improve liver damage. The direct target discovery approach – based on human genome level data – aims to identify high-impact FLD variants. This strategy may lead to a progressive improvement in risk stratification, coupling the information carried by the characterisation of risk variants with information derived from several bioinformatic “omics” approaches. The following characterisation of the disease mechanisms in experimental models leads to novel therapeutics. Among all the efforts devoted to the pursuit of a personalised medicine approach, this represents an optimal strategy for the modulation of disease pathways during pre-clinical studies and then in clinical trials. The example of PNPLA3 I148M variant is illustrated in the outer circle of the figure (in green). The modulation of genes involved in lipid droplet remodelling and lipotoxicity employing ASOs, as in the case of PNPLA3 risk variant, may be a successful blueprint. ASO, anti-sense oligonucleotides; FLD, fatty liver disease; HSD17B13, hydroxysteroid 17-beta dehydrogenase 13; PNPLA3, patatin like phospholipase domain containing 3; PRS, polygenic risk score.

The most striking example of this paradigm is being offered by the PNPLA3 I148M variant, as the main inherited and common risk factor for severe steatohepatitis, cirrhosis and HCC related to FLD. Since fat accumulation in I148M carriers is due to alterations in lipid turnover rather than DNL, therapies targeting hepatic lipogenesis have limited efficacy in mutation carriers., Conversely, directly silencing the hepatic expression of the PNPLA3 risk variant using novel therapeutics, such as oligonucleotides, may be an effective strategy (Fig. 3, outer circle in green)., This technology has already been approved to treat severe forms of dyslipidaemia and metabolic disorders. Indeed, down-modulation of expression of PNPLA3 – associated with a linked genetic variant (E434K) – has a beneficial impact on liver injury in I148M risk variant carriers and was not associated with unfavourable phenotypes. PNPLA3 silencing has been tested in vitro and in animal models. In mice, downregulation of PNPLA3 improved liver fat levels. Injection of antisense oligonucleotides against hepatic Pnpla3 in mice fed steatogenic and steatohepatitis-inducing diets reduced fat, inflammation and fibrosis, with a more pronounced benefit in animals bearing the I148M protein variant. The beneficial effect of PNPLA3 downregulation on hepatocellular fat and HSCs transactivation could also be achieved with small molecules, such as momelonib, which is active on the JAK1/2 and TGF-β/SMAD pathways. Even though this compound could lead to off-target effects and may not be suitable for chronic administration in a non-neoplastic condition, the aforementioned study provides proof-of-principle that small molecules may achieve the goal of suppressing PNPLA3. Meanwhile, phase I clinical studies have been registered, in which the safety, tolerability and pharmacokinetics/dynamics of escalating doses of liver-targeted PNPLA3-antisense oligos (e.g. NCT04142424, NCT04483947) will be tested in obese individuals and patients with NASH who are homozygous for the PNPLA3 I148M variant.

Another potential approach to counteract the detrimental impact of the PNPLA3 I148M variant may be represented by the inhibition of HSD17B13. However, HSD17B13 activity remains undefined. While the protective genetic variant has been associated with downregulation of hepatic PNPLA3 expression, it may involve regulation of retinol metabolism, with potentially widespread biological effects. In addition, experimental evidence related to PNPLA3 and MERTK variants suggests that impaired retinol metabolism may be involved in triggering inflammation and fibrogenesis during steatohepatitis.,,,[78], [79], [80]

It is worth noting that enrolment or stratification based on PNPLA3 I148M, the major genetic risk variants for FLD, and ethnicity, may improve the outcomes of clinical trials in steatohepatitis, independently of the drug target. Indeed, this would enable the selection of a more homogeneous subset of patients in terms of pathophysiology and disease progression. On the other hand, given the different representation of genetic risk variants, targeted drugs may be more effective in specific ethnic groups, e.g. individuals of Hispanic rather than African ancestry for PNPLA3-targeted drugs.

Promising therapeutic targets that have emerged from genetic studies are summarised in Table 1. Owing to their novelty, we will focus on data emerging during the last year on the role of lipotoxicity and in particular of MBOAT7 and interleukin-32 (IL-32) in the pathogenesis of FLD, with a final note on the gut-liver axis and bile acid metabolism. However, other promising targets include a) modulation of cholesterol metabolism, which is associated with the induction of steatohepatitis, particularly in patients negative for the PNPLA3 variant ; b) oxidative and endoplasmic reticulum (ER) stress, whose involvement in steatohepatitis and progressive liver disease is supported by several genetic loci, likely including the latest protective association detected at MARC1.

The possible beneficial impact of drugs targeting FLD genes on ASH remains to be proven, since hepatic damage is thought to be partly mediated by ethanol metabolites (e.g. acetaldehyde).

From quantitative to qualitative alterations of liver fat: The case of MBOAT7 and LPI

Thus, human genetics and complementary epidemiological evidence are consistent with the notion that hepatic fat accumulation plays a causal role in progressive liver disease.,, However, initial data are beginning to shed light on the qualitative alterations in liver fat (from lipidomics) and activation of intracellular pathways (from transcriptomics) that come with worsening FLD and are involved in disease progression. The most notable recent example is provided by the identification of the rs641738 C>T variant of MBOAT7 as a risk factor for FLD,,, and the discovery of the mechanism underlying the association.,, Indeed, this discovery pinpointed a role for a specific lipid species (namely lysophosphatidyl-inositol LPI), revealing new research avenues.

The rs641738 C>T MBOAT7 variant increases the risk of the full spectrum of FLD, but it was first identified as a genetic determinant of the susceptibility towards alcohol-related cirrhosis,,, and is a general modifier of liver disease progression., The impact on the full spectrum of FLD, from hepatic fat accumulation, to NASH, fibrosis and HCC has been confirmed by a recent large meta-analysis considering over a million individuals of various ethnicities. This risk allele has also been associated with an increased risk of HCC in patients with NAFLD without advanced fibrosis, and an additional rare likely pathogenic variant has recently been detected in NAFLD-HCC.

MBOAT7 encodes LPI acyltransferase 1 (also known as LPIAT1). It is an ER membrane protein with 6 transmembrane domains. MBOAT7 is involved in phospholipid acyl-chain remodelling in the so-called Land’s cycle, incorporating arachidonic acid (AA) and other polyunsaturated fatty acids (PUFAs) into LPI and other lysophospholipids. Indeed, biallelic loss-of-function mutations in MBOAT7 cause an early onset and severe neurological phenotype with cognitive impairment. This phenotype is associated with impaired neuronal and myelin development, and fully recapitulated in Mboat7 knock-out mice, which showed a marked deficit of incorporation of AA into LPI., MBOAT7 belongs to a family of membrane-bound acyltransferases that catalyse the transfer of acyl-CoA to several lipid substrates. The catalytically active site is in the ER lumen and comprises a conserved asparagine and a preserved histidine at position 321 and 356 of the protein.,

In human hepatocytes, the rs641738 C>T variant causes the downregulation of hepatic MBOAT7 both at the level of mRNA expression and protein synthesis (about 50% lower in carriers of the risk allele, who make up more than one-third of the general population),, possibly due to linkage with variation of the MBOAT7 3’-untranslated region. In keeping with the main enzymatic activity of MBOAT7, the result is impaired remodelling of plasmatic and hepatic phosphatidylinositol (PI) species, and in particular the reduction of AA-containing PI, in patients carrying the risk variant, without major changes in the composition of other phospholipids.,,,, This phenotype was fully recapitulated by knocking out MBOAT7 in experimental models in human hepatocytes and in mouse livers.,[90], [91], [92]

However, the relevance of MBOAT7 downregulation is not limited to carriage of the rs641738 risk variant. Importantly, downregulation of hepatic MBOAT7 is also associated with liver damage and adiposity-insulin resistance independently of the genetic background., Furthermore, MBOAT7 downregulation during insulin resistance was also observed in the adipose tissue,, and confirmed in animal and in vitro models. Hepatic Mboat7 transcription was curtailed in response to the rise in circulating insulin and the activation of Akt-dependent insulin signalling in mice, in response to refeeding and insulin injection, and in primary hepatocytes.

MBOAT7 downregulation has a causal role in the pathogenesis of FLD. Indeed, the silencing of hepatic MBOAT7 or hepatocellular specific deletion led to fat accumulation in hepatocytes both in vivo and in vitro.,[90], [91], [92] The impact was comparable to the effect of the rs641738 genetic risk variant. The resultant impairment in LPI metabolism led to a reduction of AA-containing PI and the accumulation of saturated LPI, which is converted to TAG by the alternative pathway through diacylglycerol, contributing to LD formation (Fig. 4)., This process is associated with upregulation of lipogenesis with SREBP-1c., Moreover, the deficiency of AA-containing PI upregulated CDP-diacylglycerol synthase, causing accelerated PI synthesis, and promoted PI degradation into diacylglycerol by phospholipase C, triggering a vicious cycle that generates TAG leading to steatosis. All in all, these new discoveries suggest that MBOAT7 downregulation represents a physiological mechanism regulated by insulin that accompanies hepatic DNL in post-prandial conditions, facilitating the incorporation of fatty acids into TAG and LD in a non-toxic form. However, during chronic hyperinsulinemia related to insulin resistance this process may become maladaptive, because it sustains hepatic lipogenesis leading to FLD, in particular in carriers of the MBOAT7 risk variant who experience more severe enzymatic deficiency. Future studies are warranted to examine whether replenishing AA-containing PI may attenuate liver damage in experimental models and in patients with dysmetabolism (See Fig. 4).

Fig. 4

MBOAT7 downregulation and LPI accumulation leads to steatohepatitis.

MBOAT7 localises to the endoplasmic reticulum, where it catalyses the transfer of AA to LPI resulting in AA-PI, that is incorporated into cell membranes. In carriers of the rs641738 risk variant or during hyperinsulinemia, MBOAT7 is downregulated, resulting in a reduction of AA-PI and accumulation of LPI that is converted into TAG by the alternative synthesis pathway through the intermediate DAG, a process promoted by SREBP1c. Deficiency of AA-PI upregulates CDS, enhancing LPI synthesis. In HSCs, LPI binds to GPR55 triggering inflammatory and profibrotic gene expression and ECM deposition. AA, arachidonic acid; AA-PI, AA-containing PI; CDS, CDP-diacylglycerol-synthase; DAG, diacylglycerol; ECM, extracellular matrix; HSCs, hepatic stellate cells; LPI, lysophosphatidyl-inositol; MBOAT7, membrane bound O-acyltransferase domain containing 7; PI, phosphatidylinositol; SREBP-1c, sterol regulatory element-binding protein 1c; TAG, triacylglycerol.

Of note, increasing the hepatocellular fat content may not be the only mechanism by which MBOAT7 downregulation promotes liver disease. This hypothesis is supported by the strong association between MBOAT7 and hepatic fibrosis that has emerged from human genetic data.,,,,, Experimental models confirmed that downregulation of MBOAT7 promoted hepatic fibrogenesis. In hepatic spheroid models composed of hepatocytes and HSCs, downregulation of MBOAT7 induced profibrotic and proinflammatory cytokines, leading to the activation of HSCs, to the expression of fibrogenic genes and to collagen accumulation. Similarly, genes implicated in inflammation and fibrosis were overexpressed in Mboat7-deficient mice, which exhibited hepatic fat accumulation and impaired LPI remodelling, and were overall more susceptible to fibrotic NASH.,[90], [91], [92] Although carriage of the MBOAT7 risk variant is associated with histological inflammation on liver biopsy in patients at risk of NASH, and hepatic Mboat7 downregulation may facilitate lipotoxicity-induced acute inflammation in experimental models, some evidence suggests that MBOAT7 downregulation can trigger hepatic fibrosis independently of inflammation.,, A possible mechanism of liver injury and fibrogenesis might be related to the accumulation of LPI. Indeed, circulating levels of LPI are increased in patients with advanced liver fibrosis compared to healthy controls, while AA-containing PI is reduced. Moreover, in high-fat diet-fed mice, LPI administration induced inflammatory and fibrotic genes, especially when Mboat7 was downregulated.

Recent evidence implicated the G protein-coupled receptor 55 (GPR55) in the pathogenesis of NASH. GPR55 is a cannabinoid receptor, whose main endogenous ligand is LPI. Fondevila et al. found that LPI increased GPR55 expression and promoted hepatic lipid accumulation in in vitro and in vivo models; moreover, GPR55 was overexpressed in patients with FLD and NASH. Hepatic injury induced by LPI seemed to be partly mediated by GPR55, since the inhibition of the receptor reduced hepatic lipid content and profibrotic gene expression in mice. Accordingly, activation of the LPI/GPR55 axis increased the expression of profibrotic genes and extracellular matrix production in HSCs, and stimulated cell proliferation in a GPR55-dependent fashion in vitro. Therefore, MBOAT7 downregulation may directly promote liver damage by increasing LPI concentration and inducing liver fibrosis via GPR55 (Fig. 4). Downregulation or antagonism of hepatic GPR55 may therefore represent another potential therapeutic strategy for steatohepatitis.

Lipotoxicity and IL-32 at the interface between fat accumulation and progressive liver disease

While the factors modulating the fate of lipids in the liver during steatohepatitis are only beginning to be understood, there is already robust evidence that dysmetabolism and insulin resistance lead to lipotoxicity and inflammation, contributing to liver damage, and representing the main trigger (together with excess alcohol) for the phenotypic expression of genetic risk variants of FLD. Incidentally, these genetic epidemiology data are consistent with the major impact of weight loss and the possible utility of anti-obesity drugs in the treatment of NASH. Besides acting as an energy store, adipose tissue also exerts its function as an endocrine organ by releasing cytokines, hormones and growth factors. Collectively defined as adipokines, these mediators are required for the regulation of metabolism, immune response and homeostasis at a systemic level. Excess adiposity and the subsequent infiltration of adipose tissue with immune cells, ultimately result in altered adipokines, promoting metabolic complications and liver damage. Furthermore, chronic inflammation and insulin resistance induce FFA release from the adipose tissue, facilitating hepatocellular fat accumulation, while inflammatory mediators drive the deposition of fibrotic tissue, the hallmark of the transition to severe liver disease.

Among the plethora of inflammatory markers associated with obesity and FLD, IL-32 is emerging as a master regulator of obesity-driven inflammation, bridging fat excess and lipotoxicity with liver damage. Indeed, circulating IL-32 levels correlate with adiposity and decrease after bariatric surgery. Similarly, IL32 mRNA levels are upregulated in the adipose tissue of obese individuals, and correlate with the enhanced expression of pro-inflammatory mediators, including IL-1ꞵ, IL-6 and IL-10, suggesting that IL-32 acts as a master regulator orchestrating obesity-driven inflammation. In human visceral adipocytes, lipopolysaccharide and tumour necrosis factor-α (TNF-α) administration induces IL-32 transcription. IL-32α increased the expression of IL-1ꞵ, TNF-α and extracellular matrix-remodelling genes, whereas IL-32 silencing yielded the opposite effects. Accordingly, serum levels of IL-32 are higher in patients with type 2 diabetes and correlate with body mass and fasting blood sugar.

IL-32 is also highly expressed in the liver and induced during liver disease., We recently showed that IL-32 is the most robustly upregulated transcript in obese individuals with severe NAFLD, in particular in carriers of the PNPLA3 I148M risk variant, and can be induced by lipotoxicity in hepatocytes. IL-32β was the most expressed isoform in hepatocytes, and hepatic IL-32 was coregulated with a set of inflammatory genes and chemokines. Accordingly, it was recently shown that PNPLA3 I148M induces metabolic reprogramming, leading to a shift in TAG/phospholipid composition and activating inflammatory pathways. Moreover, in individuals without inflammatory diseases, circulating IL-32 correlated with the hepatic transcript, and was higher in patients with severe NAFLD, improving the diagnostic accuracy of non-invasive biomarkers. Dali-Youcef et al. reported that hepatic IL-32 was upregulated in isolated steatosis and even more so in NASH, which was paralleled by an increase in inflammatory cytokines. Furthermore, IL-32 abrogated insulin-dependent AKT phosphorylation in primary human hepatocytes, implying that IL-32 may be causally involved in insulin resistance, and it may also facilitate intracellular lipid accumulation by modulating cholesterol efflux.

Notably, IL-32 and TNF-α mutually increase the transcription of each other, triggering a positive feedback loop that drives low-grade inflammation in lipid-storing organs by inducing IL-1ꞵ, IL-6 and IL-10 secretion. The IL32 promoter contains binding sites for fatty acid-responsive transcription factors, consistent with upregulation during lipotoxicity ; IL-32 has been implicated in triggering endothelial inflammation in response to a post-prandial increase in FFAs. Mechanistically, IL-32α was shown to promote STAT3 localisation at the IL6 promoter by mediating STAT3 phosphorylation by PKCε. Hepatic STAT3 signalling is enhanced in patients carrying the PNPLA3 I148M variant and in mice overexpressing mutant PNPLA3. The role of STAT3 in insulin resistance is well established: STAT3 activates SOCS3 transcription, which negatively regulates insulin signalling by interacting with the insulin receptor and insulin receptor substrate-1. Moreover, both IL-6 and ceramides amplify binding of STAT3 to the hepcidin promoter, leading to dysregulation of iron metabolism, an additional contributor to liver disease progression.

Overall, IL-32 levels correlate with obesity, insulin resistance, and steatohepatitis, suggesting that it contributes directly and indirectly to liver damage by bridging excessive intracellular fat levels with the chronic inflammation underlying these conditions. A working model is shown in Fig. 5. This hypothesis needs to be proven in experimental models of NASH, which is rendered more difficult by the fact that mice do not bear any IL32 gene orthologue. In addition, IL-32β, which is mostly expressed in hepatocytes, is not secreted through canonical pathways; therefore, the mechanism leading to the release of IL-32 into the extracellular space, as well as the signalling pathways activated by this atypical cytokine, are yet to be clarified. Circulating IL-32 may thus represent a new liver disease biomarker and a therapeutic candidate, which can be targeted by hepatic gene silencing or neutralisation by monoclonal antibodies. However, no information is yet available on IL-32 expression during ASH.

Fig. 5

Potential mechanisms linking IL-32 with liver damage in fatty liver disease and steatohepatitis.

IL-32 levels are elevated in obese and diabetic patients, suggesting a link between this cytokine and low-grade chronic inflammation and insulin resistance. More importantly, IL-32 is highly induced during liver damage, and its levels correlate with disease severity, progression to NASH and carriage of the PNPLA3 I148M variant. Acting as a master regulator of other pro-inflammatory cytokines, IL-32 promotes inflammation and liver damage by increasing the secretion of IL-1β, IL-6 and IL-10, and may also drive hepatic fibrogenesis by promoting the transcription of ECM-remodelling genes. Moreover, the IL-32 promoter is targeted by fatty acid-responsive transcription factors and thus steatosis promotes non-canonical IL-32β secretion in the liver. Both IL-32 and carriage of PNPLA3 I148M upregulate STAT3, leading to lipotoxicity and ultimately liver damage. ECM, extracellular matrix; IL-interleukin-; NASH, non-alcoholic steatohepatitis; PNPLA3, patatin like phospholipase domain containing 3.

The gut-liver axis and bile acid metabolism

Finally, candidate genetic studies are beginning to provide independent validation of the role of the gut-liver axis and bile acid metabolism in NASH. Here, the interaction between the host and microbiome genes should be considered. Indeed, both ASH and NASH are associated with altered microbial composition,, albeit with some differences., This process leads to dysbiosis and over-representation of pathogenic bacteria and metabolites, causing mucosal inflammation and alterations to the gut-vascular barrier., This alteration of the microbiome is accompanied by altered remodelling of bile acids, with a reduction of secondary bile acids that are more potent farnesoid X receptor (FXR) agonists. In experimental models, FXR agonism reversed gut-vascular barrier disruption during high-fat diet feeding, thereby protecting against NASH.

Although evidence is still at an early stage, genetic variants that modulate the gut-liver axis recapitulate the impact of drugs that modulate this pathway. First, genetic variation at the NR1H4 locus encoding FXR (rs35724) was linked to increased expression of hepatic FXR and of key targets involved in bile acid signalling via fibroblast growth factor receptor-19 (FGF19). FGF19 – the main enterokine released by the intestine in response to FXR – regulates bile acid metabolism in the liver by binding to FGF receptor-4 (FGFR4) and the β-klotho (KLB) coreceptor. The rs35724 variant was associated with increased FGFR4 and target engagement on sterol synthesis, namely cytochrome P39a1 (CYP39A1). Indeed, during steatohepatitis, reduced levels of FXR and FGF19 can lead to the accumulation of bile acids in the intestine and the liver, where their synthesis from cholesterol is not inhibited and export to the bile is downregulated, thereby promoting inflammation and carcinogenesis.

In keeping with the beneficial impact of FXR agonists on liver fibrosis in patients with NASH, the gain-of-function NR1H4 variant was associated with protection against severe NAFLD and fibrosis. Vice versa, a genetic variant in KLB associated with reduced protein expression sensitised obese children to liver damage. Strikingly, genetic predisposition to increased FXR activity also recapitulated the main adverse effect of both FXR and FGF19 agonists, namely the increase in circulating cholesterol – worsening the cardiovascular risk profile.,, The highlighted underlying mechanism, i.e. increased cholesterol synthesis, also provided the rationale for the beneficial and possibly synergic impact of co-treatment with statins.,

The role of this “FXR-FGF19-KLB-bile acids-microbiome” pathway in the pathogenesis of NASH is presented in Fig. 6. The recent demonstration that the presence of FXR is necessary for modulation of the bile acid pool and consequently of the intestinal microbiome in response to FXR, and that modulation of specific bacterial species may predict the beneficial impact of therapy with FGF19 agonists on liver damage, collectively suggest that these processes are best described as an interconnected cycle (See Fig. 6).

Fig. 6

Human genetics supports the involvement of the gut-liver axis in NASH.

NASH is associated to microbiota dysbiosis, and the over representation of pathogenic bacterial species leads to impairment in the gut-vascular barrier and bile acid management. Reduction in secondary bile acids results in hampered FXR signalling, facilitating intestinal inflammation and further impairment of the gut-blood barrier, which can be rescued by FXR agonists. FXR activation results in the release of FGF19, an enterokine that regulates hepatic bile acid metabolism by interacting with FGFR4 receptor and KLB co-receptor. Reduction of FXR in steatohepatitis negatively affects bile acid metabolism, and therefore bile acids accumulate in the liver promoting inflammation and carcinogenesis. Notably, rs35724 falling in the NR1H4 locus is protective against liver damage, but increases cardiovascular risk by increasing cholesterol synthesis; on the other hand, a KLB variant reducing protein levels promotes liver damage in obese patients. Pathways altered during NASH are indicated by red arrows, with potential therapeutic approaches in green. BAs, bile acids; CHOL, cholesterol; FGF, fibroblast growth factor; FGFR, FGF receptor; FMT, faecal microbiota transplant; FXR, farnesoid X receptor; HCC, hepatocellular carcinoma; KLB, klotho beta; NASH, non-alcoholic steatohepatitis; SCFAs, short-chain fatty acids.

Conclusions

In summary, human genetics offers a new approach to the development of therapeutics for ASH and NASH that may have predicted the outcomes, pinpointing both the strengths and weaknesses, of drug approaches undergoing evaluation in clinical trials. Furthermore, molecular genetics is highlighting new drug targets and has already led to the design of the first ever precision medicine approaches for a common liver disease, which are entering the clinical research arena. These aim at targeting the genetic risk variants to cure steatohepatitis by suppressing a contributing cause. Given the common genetic pathophysiology of ASH and NASH, there is also the potential and hope that some of the novel approaches developed for NASH can be repurposed to treat AFLD and ASH, but additional studies are needed to prove this hypothesis. Most importantly, this class of drugs may be particularly effective in patients with FLD and steatohepatitis,, with dysmetabolism associated with moderate alcohol consumption, who cannot be diagnosed with NASH, but who are at high risk of disease progression.

Financial support

MyFirst Grant n.16888, Ricerca Finalizzata Ministero della Salute RF-2016-02364358, , the European Union (EU) Programme Programme “Photonics” under grant agreement “101016726” for the project “REVEAL: Neuronal microscopy for cell behavioural examination and manipulation”, the Innovative Medicines Initiative 2 joint undertaking of European Union’s research and innovation programme and EFPIA for the project LITMUS- “Liver Investigation: Testing Marker Utility in Steatohepatitis” under grant agreement No. 777377), “Liver BIBLE” PR-0391, core COVID-19 Biobank (RC100017A) to LV.

Authors’ contributions

LV study conception design, CB, EC, and LV: first manuscript drafting all authors were involved in manuscript writing and revision.

Conflict of interest

LV has received speaking fees from MSD, Gilead, AlfaSigma and AbbVie, served as a consultant for Gilead, Pfizer, Astra Zeneca, Novo Nordisk, Intercept, Diatech Pharmacogenetics and Ionis Pharmaceuticals, and received research grants from Gilead.

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