Liver cell apoptosis is a prominent feature associated with the severity of NASH (Feldstein et al., 2003). Apoptosis, a process of programmed cell death, is a method of cellular self-destruction that removes damaged cells to maintain homeostasis under both normal and pathophysiological conditions. In the pathogenesis of NASH, hepatocellular apoptosis is recognized as an important mechanism that can contribute to liver fibrosis. FFAs mediate lipoapoptosis, which is a potential mechanism of apoptosis associated with NASH (Malhi et al., 2006). Long-chain fatty acids (LCFAs) enter hepatocytes through FAT/CD36, originally characterized as an FA transport protein. These FAs can directly contribute to triacylglycerol accumulation in hepatocytes. FFAs can also enter the mitochondria and promote the activity of electron transport chain and FA oxidation, resulting in increased ROS production. Excessive ROS occur oxidative stress and alter cellular biomolecules (lipids, proteins, DNA), leading to mitochondrial dysfunction and liver injury (Evans et al., 2002; Cazanave and Gores, 2010). The c-Jun N-terminal kinase (JNK, also known as stress-activated protein kinase) signaling pathway is a key regulator in ROS-mediated cell death that well describes the relationship between elevated FFAs and lipoapoptosis as features of NAFLD (Malhi et al., 2006; Amir et al., 2012). FFAs can elevate ROS/oxidative stress and sequentially stimulate JNK activation. Activated JNK induces the lipoapoptosis of hepatocytes by stimulating the pro-apoptotic Bcl-2 protein Bax, which triggers the mitochondrial apoptosis pathway (Malhi et al., 2006). There are two major signaling pathways leading to apoptosis: the intrinsic apoptosis pathway initiated by mitochondrial events and the extrinsic apoptosis pathway initiated by death receptors such as tumor necrosis factor (TNF), TRAIL, and FAS-L (Dhanasekaran and Reddy, 2008). JNK can play a central role in both apoptosis pathways (Dhanasekaran and Reddy, 2008). The endoplasmic reticulum (ER) is also an organelle closely associated with lipotoxicity and lipoapoptosis (Cazanave and Gores, 2010). Both aberrant FFAs and FFA-induced ROS are critical factors that can trigger ER stress. Sustained ER-stress can activate the JNK signaling pathway and sequentially modulate the pro-apoptotic Bcl-2 family members, leading to hepatocyte apoptosis.

Conversely, AMP-activated protein kinase (AMPK), a mitochondria fine-tuning factor, can also be stimulated by ROS. Mitochondria-derived ROS acts on redox-sensitive cysteine residues (Cys-299/Cys-304) on the AMPKα subunit, and elevated ROS can activate AMPK by decreasing ATP levels or through S-glutathionylation of cysteine in the AMPKα and AMPKβ subunits (Filomeni et al., 2015; Hinchy et al., 2018). Activated AMPK occurs a PGC1α-dependent antioxidant response and lowers mitochondria ROS production, allowing for the maintenance of metabolic homeostasis and survival (Rabinovitch et al., 2017). AMPK exerts protective effects against hepatocyte lipotoxicity in the AMPK–PGC1α and AMPK–sirtuin 1 (SIRT1) axes (Li et al., 2020; Fang et al., 2022). Indeed, AMPK activity is reduced in NAFLD and NASH, and liver-specific AMPK reduction leads to NASH phenotypes such as fibrosis and cell death (Li et al., 2020; Fang et al., 2022). AMPK can ameliorate liver fat accumulation and NASH-associated hepatocyte apoptosis (Li et al., 2020; Fang et al., 2022).

DAMPs, also termed as alarmins, are endogenous risk factors that are released from damaged or dying cells to trigger strong inflammatory responses by activating the innate immune system via interactions with pattern recognition receptors (Bianchi, 2007; Roh and Sohn, 2018). Injured hepatocytes also release DAMPs, including metabolites, microRNAs, mitochondrial DNA, and mitochondrial double-stranded RNA (Kubes and Mehal, 2012). These DAMPs are recognized by macrophages, and they stimulate multiple inflammatory pathways through Toll-like receptors (TLRs) and inflammasomes (Zhang and Mosser, 2008; Schaefer, 2014). In NASH progression, hepatocyte-derived DAMPs promote the activation of Kupffer cells to remove damaged hepatocytes and accelerate liver inflammation (Kazankov et al., 2019). Activated Kupffer cells stimulate HSC activation by secreting inflammatory cytokines [TNFα, interleukin-1β (IL-1β), IL-6, C-C motif chemokine ligand 5 (CCL5)] and producing pro-fibrotic mediators (transforming growth factor β (TGF-β), platelet-derived growth factor [PDGF], connective tissue growth factor (CTGF)] (Wen et al., 2021). Additionally, HSCs have also a purinergic receptor P2Y14 as the DAMP receptor. The P2Y14 ligands uridine 5′-diphosphate (UDP)-glucose and UDP-galactose are abundant in hepatocytes. HSCs can be activated by P2Y14 ligands or secretory factors from dying hepatocytes in a P2Y14-dependent manner (Mederacke et al., 2022). Namely, DAMPs from damaged hepatocytes can promote inflammation and fibrosis by stimulating Kupffer cells and HSCs during NASH progression.

Autophagy is a self-degradative process and cellular recycling system that allows cells to degrade unnecessary or damaged organelles and pathogens to balance energy sources in response to environmental and nutrient stresses (Glick et al., 2010). Autophagic processes are directly linked with the development and progression of NAFLD. In the pathogenesis of NAFLD, the autophagy pathway mediates the degradation of intracellular lipids in hepatocytes, suggesting that the autophagy pathway is closely involved in the development of hepatic steatosis (Czaja, 2016). Simple steatosis (also known as NAFL) progresses to NASH, which is characterized by hepatocyte injury and death, inflammation, and fibrosis associated with increased oxidative stress and inflammatory cytokines (Browning and Horton, 2004; Chalasani et al., 2004). Autophagy regulates cell death signaling pathways induced by oxidants and TNF, which mediate NASH injury, and protects against cellular injury by removing damaged organelles in NASH (Czaja, 2016). NAFLD impairs autophagy, which is believed to involve various signaling pathways. Alterations in AMPK and mTORC1 activity can lead to impaired autophagy during NAFLD (Garcia et al., 2019; Li et al., 2019). Upregulation of CD36, a facilitator of membrane FA transport, inhibits autophagy initiation in the fatty liver by inhibiting the AMPK/uncoordinated 51-like kinase 1 (ULK1)/Beclin1 pathway. The negative effect of CD36 on autophagy is caused by the suppression of AMPK, a major activator of autophagy (Garcia et al., 2019; Li et al., 2019). Conversely, CD36 deficiency increases autophagy by activating AMPK, which enhances the activity of ULK1 and Beclin1that is important for autophagosome biogenesis (Mercer et al., 2018; Li et al., 2019). mTORC1 activation in NAFLD can impair autophagy by inhibiting ULK1/2 and inactivating the autophagy enhancer (Pacer) protein, thus interfering with autophagosome–lysosome fusion (Chen and Kilonsky, 2011; Carroll and Dunlop, 2017; He et al., 2020). mTORC1 activity might be influenced by AMPK (Carroll and Dunlop, 2017; He et al., 2020).

Intracellular Ca²⁺ homeostasis is important for regulating lipid and carbohydrate metabolism in normal hepatocytes (Ali et al., 2021). Chronic lipid exposure can alter Ca²⁺ homeostasis in hepatocytes (Ali et al., 2019; Ali et al., 2021). Lipids increase Ca²⁺ efflux from the ER and consequentially increase Ca²⁺ concentrations in the cytoplasm and mitochondria. Lipids can reduce lipolysis and β-oxidation and enhance lipogenesis, ER stress, ROS generation, and Ca²⁺/calmodulin-dependent kinase activation, resulting in the progression of NAFLD to more severe forms such as NASH, fibrosis, and HCC (Ali et al., 2019). Ca²⁺ is a key regulator of autophagy that acts through the regulation of pathways such as rapamycin complex 1, calcium/calmodulin-dependent protein kinase II, and protein kinase C signaling (Bootman et al., 2018). Ca²⁺ is also involved in autophagic signaling pathways including both mTOR and AMPK (Bootman et al., 2018). During obesity involving lipotoxicity, cytoplasmic Ca²⁺ levels are abnormally increased, and this is believed to attenuate autophagic flux by inhibiting autophagosome–lysosome fusion (Park et al., 2014). Obese mice treated with the calcium channel blocker verapamil exhibited rescued cytoplasmic Ca²⁺ levels in hepatocytes and impaired autophagosome–lysosome fusion (Park et al., 2014). These results indicate that altered intracellular Ca²⁺ levels can affect autophagy signaling during NAFLD progression.

EMT is the cellular process in which epithelial cells lose their apical–basal polarity and junction and acquire mesenchymal features (Kalluri and Neilson, 2003; Thiery et al., 2009). EMT contributes to organ fibrosis and promotes carcinoma progression (Thiery, 2002; Kalluri and Neilson, 2003; Thiery et al., 2009). EMT confers migratory and invasive properties to cells (Thiery et al., 2009). Because of cellular plasticity in the liver, epithelial cells can acquire fibroblastic properties through EMT, leading to fibrosis. Liver fibrosis is a protective reaction against chronic liver damage from multiple etiologies (Lotersztajn et al., 2005). It is characterized by the excessive deposition of extracellular matrix (ECM) produced by myofibroblasts (Miao et al., 2021). Activated HSCs (aHSCs) are considered the major origin of myofibroblasts (Iwaisako et al., 2012). However, hepatocytes can also undergo EMT to influence the fibroblastic phenotype during liver fibrosis (Zeisberg et al., 2007; Nitta et al., 2008). Hepatocytes stimulated with TGF-β1 display fibroblast-like morphology and express fibroblast-specific protein 1, which accounts for EMT in hepatocytes (Zeisberg et al., 2007). Other studies also showed that hepatocytes undergo EMT-like phenotypic changes in TGF-β-dependent manner and participate in fibrogenesis (Kaimori et al., 2007; Kojima et al., 2008). Conversely, specific inhibition of TGF-β signaling in hepatocytes delayed the fibrogenic phenotype (Dooley et al., 2008). The anti-fibrotic protein apamin can inhibit hepatic fibrogenesis by interfering with TGF-β1–induced EMT in hepatocyte (Lee et al., 2014). Celecoxib, a selective cyclooxygenase-2 (COX-2) inhibitor, ameliorates hepatic fibrosis by inhibiting hepatocyte EMT (Wen et al., 2014). These findings indicate that hepatocyte EMT contributes significantly to liver fibrosis.

Hippo signaling plays an important role in the control of regeneration, stem cell self-renewal, and liver size (Pibiri and Simbula, 2022). It was recently reported that the Hippo signaling pathway is closely associated with fibrosis induction in multiple organs (Mia and Singh, 2022). Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ), are the downstream effects of the mammalian Hippo pathway, and they function as transcriptional co-activators (Luo and Li, 2022). Recently, it was reported that YAP and TAZ in hepatocytes are hyperactivated during liver injury (Mia and Singh, 2022). YAP-expressing hepatocytes strongly induce genes related to inflammation (TNF, IL-1β) and fibrosis [collagen type I alpha 1 chain (COL1A1), tissue inhibitor of metalloproteinase 1 (TIMP-1), PDGF-C, TGF-β2] (Mia and Singh, 2022). Conversely, hepatocyte-specific deficiency of YAP and TAZ results in reduced inflammation, myoblast expansion, and liver fibrosis (Mia and Singh, 2022). Excessive cholesterol in hepatocytes can also promote liver fibrosis by upregulating TAZ through the Ca²⁺- RhoA signaling pathway (Wang et al., 2020). It has been reported that hepatocyte-targeted TAZ silencing in murine NASH models improves hepatic inflammation, and suppresses hepatocyte cell death and fibrosis (Wang et al., 2016). These results indicate that the Hippo signaling pathway in hepatocytes is involved in inflammatory and fibrotic processes during NAFLD/NASH progression.

Cholangiocytes are morphologically and functionally heterogeneous (Glaser et al., 2009; Maroni et al., 2015). In humans, cholangiocytes are divided into small, medium, and large cells based on the correspondence with the diameter of the bile duct (BD) (Kanno et al., 2000; Glaser et al., 2009; Franchitto et al., 2013). In rodents, they exist in small and large sizes (Tabibian et al., 2013; Maroni et al., 2015; Little et al., 2023). Cholangiocytes exhibit different functions depending on their size. Small cholangiocytes line along the small intrahepatic BDs and have a relatively undifferentiated phenotype with large nuclei and small cytoplasm (Franchitto et al., 2013; Sato et al., 2018). They reside in the biliary progenitor cell compartment of the liver. They are known to be more resistant to damage than large cholangiocytes, and when large cholangiocytes are damaged by severe injury, they can differentiate into large cholangiocytes via Ca²⁺-activated signaling (Mancinelli et al., 2010; Mancinelli et al., 2013). Namely, small cholangiocytes can function as progenitor cells for large cholangiocytes. Large cholangiocytes line along the large intrahepatic and extrahepatic BDs and have small cytoplasm and large nuclei (Franchitto et al., 2013; Sato et al., 2018). Large cholangiocytes are thought to be more mature cholangiocytes than small cholangiocytes (Maroni et al., 2015). They are dependent on cyclic adenosine monophosphate (cAMP) signaling and are sensitive to injury (Glaser et al., 2009; Maroni et al., 2015). During liver injury, large cholangiocytes can undergo senescence and enter a senescence-associated secretory phenotype (SASP) state, secreting proinflammatory factors that exacerbate the damage (Huda et al., 2019). On the other hand, when the large cholangiocytes are damaged, small cholangiocytes proliferate and transdifferentiate into a large cholangiocyte phenotype, playing an important role in repairing the damaged epithelial lining (Mancinelli et al., 2013). These cell size-dependent differential responses will be thought to help understand the heterogeneity of cholangiocytes.

Under the healthy condition, quiescent cholangiocytes function as immunological and antimicrobial defense systems by secreting biliary immunoglobulin (especially secretory IgA) and various antimicrobial factors such as β-defensin 2, mucins, and lactoferrin into the bile (Farina et al., 2011; Demetris et al., 2016). These immunological functions are closely linked with the intestinal mucosal immune system (Chen et al., 2008; Demetris et al., 2016). Cholangiocytes that play an active role in immune pathogenesis fundamentally express TLRs (Chen et al., 2005; Chen et al., 2008). TLRs expressed by cholangiocytes recognize pathogen-associated molecular patterns (PAMPs) derived from the intestine or bloodstream, and consequently activated downstream signals induce the secretion of antimicrobial factors and inflammatory cytokines (Chen et al., 2005; Ninlawan et al., 2010; Oya et al., 2014). The complicated interaction between innate and adaptive immunity not only enhances biliary defense but also can promote inflammation. For example, intestine-derived lipopolysaccharide (LPS), a representative of typical PAMPs, and hepatocyte-derived DAMP can bind to TLR4 expressed by biliary epithelial cells during NAFLD development and progression (Guo and Firedman, 2010; Fabris et al., 2017). Persistent liver damage activates cholangiocytes to participate in liver inflammation by releasing inflammatory cytokines (Pinto et al., 2018).

Cholangiocytes are antigen-presenting cells that fundamentally express human leukocyte antigen (HLA) class I molecules (Syal et al., 2012; Schrumpf et al., 2015). Adhesion molecules such as HLA class II, leukocyte functioning antigen (LFA)-3, and intercellular adhesion molecule-1 (ICAM-1) are also expressed in cholangiocytes (Syal et al., 2012). Unconventional T cells are densely populated in sections adjacent to the liver and intestine (Godfrey et al., 2015). Cholangiocytes can present antigens to unconventional T cells including natural killer T (NKT) cells and mucosa-associated invariant T (MAIT) cells (Godfrey et al., 2015). Unconventional T-cell receptors recognize antigens expressed by HLA class I molecules (Mayassi et al., 2021). NKT cells perceive lipid antigens by a cluster of differentiation (CD) 1 by cholangiocytes (Schrumpf et al., 2015). MAIT cells perceive antigen bacterial B vitamins by MR1 molecules on cholangiocytes (Jeffery et al., 2016). Namely, cholangiocytes exposed to lipids activate NKT cells, and cholangiocytes exposed to bacteria activate MAIT cells, leading to releasing inflammatory cytokines and participating in inflammatory responses (Schrumpf et al., 2015; Jeffery et al., 2016).

Cholangiocytes can be activated by various factors such as infection, endotoxins, and FAs (O’Hara et al., 2013; O'Hara et al., 2017). Activated cholangiocytes are characterized by increased proliferation and increased secretion of proinflammatory and profibrotic factors (Pinto et al., 2018; Strazzabosco et al., 2018). Activated cholangiocytes are involved in the inflammatory responses by releasing cytokines and chemokines (Adams, 1996; Chen et al., 2008). Sustained biliary cell damage can lead to excessive scar formation, biliary cirrhosis, and the chronic proliferation of cholangiocytes (Banales et al., 2019). Signals generated by the loss of cholangiocyte homeostasis drive the repair process (Fabris et al., 2016; Banales et al., 2019). These signals are recognized by inflammatory cells including macrophages and neutrophils and scaffold-producing cells including myofibroblasts and portal fibroblasts, and endothelial cells forming vasculature (Banales et al., 2019). These cells and corresponding signals form the biliary reparative complex referred to as ductular reactions (DR) (Sato et al., 2019; Chen et al., 2022). Namely, DRs are a response to a wide diverse of hepatobiliary damage that aims to recover compromised physical function after liver injuries (Jain and Clark, 2021).

Activated cholangiocytes, inflammatory cells, and mesenchymal cells are core cells that form DRs (Banales et al., 2019). As described above, activated cholangiocytes participate in inflammatory responses by secreting proinflammatory cytokines and chemokines. These secreted inflammatory factors activate and recruit immune cells. Liver mesenchymal cells are also activated by these inflammatory signals from the bile duct and are attracted to the bile duct (Cai et al., 2023). Activated mesenchymal cells secrete vesicles and soluble paracrine factors and stimulate cells in the bile (Banales et al., 2019; Azparren-Angulo et al., 2021). This epithelial-mesenchymal interaction demands the complementary expression of agonists and their corresponding receptors by epithelial and mesenchymal cells. TGF-β, PDGF, and monocyte chemoattractant protein-1 (MCP-1)/CCL2 are known to be released by reactive ductular cells (RDCs) and can activate myofibroblast (Kinnman et al., 2003; Kruglov et al., 2006; Fabris et al., 2017). RDCs exhibit a biliary phenotype as an epithelial component located around the portal space (Roskams et al., 2004). However, they can acquire morphologically and functionally mesenchymal properties (Fabris et al., 2016). This well describes the phenotypic plasticity of RDCs. RDCs exhibit upregulated EMT markers [S100A4, vimentin, matrix metalloproteinase 2 (MMP2)] and decreased epithelial markers (E-cadherin) in chronic cholangiopathies (Fabris and Strazzabosco, 2011). This ability of RDCs to increase mobility is essential for wound repair (Park, 2012). RDCs-induced biliary repair is mediated by morphogenetic pathways (Fabris et al., 2017). Among them, the Notch and YAP/TAZ pathways play critical roles in maintaining the biliary structure during biliary repair (Fiorotto et al., 2013; Morell et al., 2017; Panciera et al., 2017). Hepatocyte Notch activation induces hepatocyte-to-cholangiocyte conversion that expresses biliary SOX9 and HNF1β (Morell et al., 2017). Direct interaction of Notch-expressing hepatic progenitor cells (HPCs) with Jagged-1-expressing portal myofibroblasts occurs HPCs-to-RDCs conversion (Strazzabosco and Fabris, 2013; Nakano et al., 2017). Additionally, Notch signaling plays an important role in branching tubulogenesis in bile duct repair (Fiorotto et al., 2013). Depletion of Notch signaling attenuates biliary repair, whereas sustained Notch activation can lead to liver epithelial dysplasia and HCC (Geisler and Strazzabosco, 2015). These suggest that reactive cholangiocytes contribute to advanced liver failure including inflammation through secretion of proinflammatory cytokines and fibrosis through cholangiocyte proliferation and direct interaction with myofibroblasts.

Circulating FAs are increased in patients with NAFLD and NASH (Puri et al., 2009; Zhou et al., 2016; Feng et al., 2017). Excessive FAs induce cholangiocyte lipoapoptosis in a FoxO3/miR-34a-dependent manner. NAFLD patients show increased DR and fibrosis as markers of cholangiocyte damage (Natarajan et al., 2014; Natarajan et al., 2017). Cytokeratin 19 (CK19) is a biliary epithelial marker that is detected in the cytoplasm of DR-positive hepatobiliary cells (de Lima et al., 2008; Cai et al., 2016). CK19-positive DRs is more pronounced in areas with severe liver damage, as confirmed in liver sections from choline-deficient high-trans-fat diet-fed rat as a NAFLD model (de Lima et al., 2008). Biliary senescence is closely related to NAFLD and NASH progression (Meijnikman et al., 2021). Senescent RDCs can amplify inflammation and fibrotic responses through the senescence-associated secretory response (SASP) mechanism. Senescent cholangiocytes exhibit increased secretion of SASP factors such as TGF-β, PDGF, TNF-α, and IL-1β (Wu et al., 2016; Lopes-Paciencia et al., 2019; Blokland et al., 2020). This induces the recruitment and infiltration of immune cells and consequently exacerbates microvesicular steatosis and fibrosis during NAFLD progression.

In the liver, TGF-β is an essential molecule for organ homeostasis that regulates organ size and growth by limiting cell proliferation and promoting apoptosis in hepatocytes (Giannelli et al., 2016; Fabregat and Caballero-Diaz, 2018). Loss of these functions can lead to hyperproliferative disorders including cancer (Dooley and ten Dijke, 2012; Giannelli et al., 2016). TGF-β exerts tumor-suppressive effects in the early stage of carcinoma, whereas it promotes carcinogenesis in the late stage of carcinoma (Massagué, 2008).

TGF-β is well-known as a key factor in HSC activation and ECM production, which triggers fibrosis in the liver (Fabregat and Caballero-Diaz, 2018). It contributes to the progression from initial liver injury, including inflammation and fibrosis, to cirrhosis and HCC (Dooley and ten Dijke, 2012). Cell plasticity can contribute to the adaptation of liver cells to metabolic stress and facilitate the transition from NAFL to NASH to fibrosis (Dooley and ten Dijke, 2012). During pathological conditions caused by chronic liver damage, HSC activation is mediated by various signals such as growth factors (e.g., PDGF, CTGF), lipids, ROS, cytokines produced by hepatocytes, Kupffer cells, endothelial cells, and cholangiocytes (Yin et al., 2013; Tsuchida and Friedman, 2017). Among these cytokines, TGF-β plays a central role in HSC activation from a quiescent state to an activated myofibroblastic phenotype (Dewidar et al., 2019). Namely, TGF-β is closely associated with the plasticity of HSCs, and it promotes liver fibrosis (Fabregat and Caballero-Diaz, 2018). At the plasma membrane level, TGF-β primarily binds to and activates types I and II serine/threonine kinase receptors to phosphorylate the downstream mediators SMAD proteins (Hata and Chen, 2016; Tzavlaki and Moustakas, 2020).

There are three different types of the SMAD proteins based on their functions: receptor-regulated SMADs (SMAD2, SMAD3, SMAD1/5/8), a common SMAD4, and inhibitory SMADs (Smad6, Smad7) (Hata and Chen, 2016). Among the SMAD proteins, TGF-β is known to upregulate SMAD2 and SMAD3, whereas downregulate SMAD7. The balance between SMAD2/3 and SMAD7 is known to play a critical role in liver fibrosis. Upon ligand binding, TGF-β receptors phosphorylate SMAD2/3, which forms a complex with SMAD4. This SMAD complex regulates the transcription of target genes including pro-fibrogenic genes along with cofactors after translocation to the nucleus (Hata and Chen, 2016). Conversely, the anti-fibrotic factor SMAD7 can directly bind to TGF-β receptors to inhibit SMAD2/3 phosphorylation (Yan et al., 2009), and can target and degrade Type I receptors by forming the complex with SMAD ubiquitination regulatory factor, an E3 ubiquitin ligase (Kavsak et al., 2000).

Several studies also describe SMAD3 as the major mediator of TGF-β–induced fibrogenic responses in HSCs, especially those associated with the induction of collagen expression (Inagaki et al., 2001; Schnabl et al., 2001; Furukawa et al., 2003). TGF-β can promote SMAD3 phosphorylation by activating p38 mitogen-activated protein kinase (MAPK), resulting in HSC activation and fibrosis progression (Inagaki et al., 2001; Schnabl et al., 2001; Furukawa et al., 2003). TGF-β and PDGF can activate HSCs via the JNK pathway (Yoshida et al., 2005). JNK in aHSCs directly phosphorylates SMAD2/3, resulting in the transcriptional activation of plasminogen activator inhibitor-1, which promotes HSC migration and fibrosis in liver tissue (Yoshida et al., 2005). Additionally, TGF-β can also activate noncanonical SMAD-independent pathways such as PI3K/AKT, mTOR, MAPK, and Rho/GTPase signaling pathways (Dewidar et al., 2019).

Meanwhile, SMAD7 can inhibit TGF-β–induced transdifferentiation and arrest HSCs in a quiescent stage by reducing the mRNA and protein levels of inhibitor of differentiation 1, which promotes hepatic fibrogenesis via the activin-like kinase 1/SMAD1 pathway (Dooley et al., 2003; Wiercinska et al., 2006). As well, miR-130a-3p can directly target TGF-β receptors and induce HSC inactivation and apoptosis, suggesting miR-130a-3p as a negative regulator in the progression of NASH through TGF-β/SMAD signaling (Wang et al., 2017).

ROS play a crucial role in the pathogenesis of liver fibrosis (Gandhi, 2012). They drive the activation of pro-fibrogenic HSCs, leading to ECM synthesis. During liver injury, ROS can be generated by two major enzyme families, namely, the NADPH oxidase (nicotinamide adenine dinucleotide phosphate oxidase [NOX]) and CYP450 families (Fabregat and Caballero-Diaz, 2018). Following liver injury, NOX isoforms in aHSCs are remarkably upregulated (Paik et al., 2011; Aoyama et al., 2012). For example, HSC-induced phagocytosis of hepatocyte-derived apoptotic bodies can activate NOX and stimulate TGF-β signaling and COL1A1 expression (Zhan et al., 2006; Jiang et al., 2010). NOX-generated ROS are also known to play essential roles in HSC activation and liver fibrosis (De Minicis and Brenner, 2007; Liang et al., 2016). ROS can stimulate TGF-β signaling by activating matrix metalloproteinases, enhancing TGF-β expression, or promoting TGF-β release (Richter and Kietzmann, 2016; Fabregat and Caballero-Diaz, 2018).

In an activated state of HSCs, NOX mediates pro-fibrogenic responses triggered by various stimuli, including TGF-β, PDGF, Ang II, and leptin (Liang et al., 2016). Among these stimuli, TGF-β is the most influential modulator that induces myofibroblastic HSC activation in the liver, leading to collagen protein and α-SMA expression (Gressner et al., 2002). The expression of NOX isoforms depends on the type of liver resident cells (Liang et al., 2016). Kupffer cells mainly express phagocytic NOX2, whereas hepatocytes, HSCs, and endothelial cells express phagocytic NOX2 and nonphagocytic NOX1 and NOX4. NOX4 mediates TGF-β-associated fibrogenic responses in various organs (Cucoranu et al., 2005; Hecker et al., 2009; Bondi et al., 2010; Sampson et al., 2011; Boudreau et al., 2012). In liver fibrosis, TGF-β enhances NOX4-induced ROS generation during HSC activation (Proell et al., 2007). In bile duct ligation (BDL)- and CCL4-induced fibrosis, TGF-β promotes NOX4 expression and activity via SMAD3 in HSCs (Jiang et al., 2012; Sancho et al., 2012).

Nuclear factor-erythroid 2-related factor 2 (Nrf2) is considered an antagonistic factor in liver fibrosis. As a transcription factor, Nrf2 modulate the expression of ROS-detoxifying enzymes such as glutathione S-transferase (GST) and glutathione peroxidase 2 (GPX2) (Tonelli et al., 2018). Nrf2 signaling represents a cytoprotective mechanism induced in cells exposed to oxidative stress (Tonelli et al., 2018). Further, Nrf2 exerts a protective effect against toxin-induced liver fibrosis (Xu et al., 2008). In line with this, sulforaphane, an Nrf2 activator, also improves hepatic fibrosis by inhibiting TGF-β signaling in the BDL model (Oh et al., 2012). Additionally, HSCs with reduced Nrf2 levels display markedly increased aHSC markers including ECM components through the TGF-β/SMAD signaling (Prestigiacomo and Suter-Dick, 2018). These well describe the association among TGF-β, ROS, and Nrf2 during the liver fibrosis process.

Hh pathway is a signaling cascade that regulates tissue morphogenesis by directing cell fate during embryogenesis and modulates injury-induced tissue remodeling in adults (Sicklick et al., 2005; Omenetti et al., 2011). It is a highly conserved and complex signaling pathway. The canonical Hh pathway consists of four main parts: 1) Hh ligands (Sonic hedgehog, Indian hedgehog, and Desert hedgehog), 2) cell surface receptor Patched (PTCH), 3) signal transducer Smoothened (SMO), and 4) glioblastoma (Gli) transcription factors as the effectors (Omenetti et al., 2011; Machado and Diehl, 2018).

HSCs participate in remodeling of the injured liver, and they can exert considerable plasticity during the fibrosis process. The Hh signaling pathway mediates adaptive responses during NAFLD, NASH, and fibrosis as well as in liver regeneration and injury (Fleig et al., 2007; Verdelho Machado and Diehl, 2016; Machado and Diehl, 2018). Hh ligands are rarely expressed in the healthy adult liver, whereas they are highly expressed during liver injury, activating the signal transduction pathway (Machado and Diehl, 2018). Hh signaling activation enhances the expression of fibrogenic genes (α-SMA, COL1A1, vimentin, and TGF-β) and Snail, a Gli-responsive transcription factor that mediates TGF-β–induced EMT, whereas it reduces the expression of qHSC markers (Choi et al., 2009). During HSC activation, bmp7 and its target id2 are downregulated, resulting in decreased E-cadherin expression (Valcourt et al., 2005). Conversely, inhibition of the Hh signaling recovers the expression of the epithelial markers (bmp7, id2, E-cadherin) and qHSC markers, and results in a loss of the fibrotic phenotype of aHSCs (Choi et al., 2009). EMT in HSC is closely associated with cytoskeletal reorganization. Therefore, it is accompanied by alterations in the activity of the small GTPase Rac1 that is associated with cytoskeleton. In both in vitro HSCs and in vivo mouse models, Rac1 activation promotes Hh signaling and the fibrogenic phenotype and aggravates liver fibrosis (Choi et al., 2010).

TLR4 is a transmembrane protein member of the TLR family that mediates the activation of innate immune responses and recognizes LPS (Kawasaki and Kawai, 2014). Obese individuals or patients with NAFLD have an overgrowth of intestinal bacteria, resulting in increased LPS production compared to the findings in normal-weight individuals (Wigg et al., 2001; Crane et al., 2015). Kupffer cells express TLR4, which recognizes LPS. Based on the gut–liver axis, intestine-derived LPS activate Kupffer cells by binding to TLR4. Activation of TLR4-mediated signaling pathway triggers large amounts of pro-inflammatory cytokines through nuclear factor-kappa B (NF-κB), MAPK, extracellular signal-regulated kinase 1, p38, JNK, and interferon regulatory factor 3 (IRF3) (Bieghs and Trautwein, 2013). Consequently, excessive pro-inflammatory cytokines can be produced during liver fibrosis, and they cause liver damage through the activation of HSCs (Bieghs and Trautwein, 2013).

TLR4 signaling consists of myeloid differentiation primary response 88 [MyD88]-dependent and MyD88-independent signaling pathways, which are mediated by various adaptor proteins. The MyD88-dependent pathway induces pro-inflammatory cytokines by directly stimulating NF-κB and activator protein 1 (AP-1) through the adaptor molecules MyD88 and MyD88 adaptor-like (Mal, also termed TIRAP) (Akira and Takeda, 2004; Guo and Friedman, 2010). On the other hand, the MyD88-independent pathway is mediated by Toll/IL-1R domain-containing adaptor-inducing IFN-β (TRIF) and TRIF adaptor-related adaptor molecule (TRAM) (Kawai et al., 2001; Doyle et al., 2002; Yamamoto et al., 2003a; Yamamoto et al., 2003b; Kagan et al., 2008; Tanimura et al., 2008). These adaptors activate IRF3, which regulates type I IFN expression, late NF-κB activation, and pro-inflammatory immune responses (Kawai et al., 2001; Doyle et al., 2002; Akira and Takeda, 2004). Activated Kupffer cells produce inflammatory factors including TNF and IL-6 and recruit infiltrating inflammatory cells. It is thought that these events contribute to NAFLD progression at multiple levels (Leroux et al., 2012). As endogenous TLR4 ligands, DAMPs derived from damaged hepatocytes can stimulate Kupffer cells (Cha et al., 2018; Takimoto et al., 2023). DAMPs can act on TLR4, which activates M1-type Kupffer cell and promote the secretion of pro-inflammatory cytokines (Takimoto et al., 2023).

ROS critically affect Kupffer cell function. The primary source of ROS in macrophages is NOX (Canton et al., 2021). Kupffer cells generate superoxide, a major form of ROS, through the phagocytic NOX2 in response to microbial stimuli, including LPS (Lambeth, 2004). It helps kill microorganisms or stimulates redox-sensitive targets including protein kinase C, NF-κB, and ERK family members (Forman and Torres, 2002).

Mitochondria represent the largest source of metabolic ROS induction (Turrens, 2003). Mitochondrial substrate oxidation creates the electrochemical proton gradient across the inner mitochondrial membrane. The electrochemical energy is used for the synthesis of ATP via oxidative phosphorylation. The energy of the electrochemical gradient can also be dispersed as heat via uncoupling. Among uncoupling proteins, uncoupling protein 2 (UCP2) is distributed in most tissue, and it is abundantly expressed in immune cells (Brand and Esteves, 2005; Mattiasson and Sullivan, 2006). UCP2 overexpression reduces ROS production and immune cell activation (Kizaki et al., 2002; Ryu et al., 2004), whereas loss of UCP2 promotes ROS production, pro-inflammatory cytokine secretion, and NF-κB activation (Arsenijevic et al., 2000; Bai et al., 2005). LPS can inhibit UCP2 expression in macrophages (Arsenijevic et al., 2000; Kizaki et al., 2002). This indicates that TLR4-mediated signaling might use mitochondrial ROS in an amplifying circuit. In fact, LPS-induced TLR4 signaling activates JNK and p38 through mitochondrial ROS in the peritoneal macrophages from ucp2^−/− mice (Emre et al., 2007). These findings suggest that UCP2 can inhibit ROS-sensitive TLR4 signaling, including JNK, p38, and NF-κB. ROS-mediated amplification of TLR4 signaling might result from dysfunctional UCP2 in Kupffer cells.

Inflammation is a key driver of liver fibrosis (Seki and Schwabe, 2015; Koyama and Brenner, 2017). During liver injury, immune cells, including macrophages, lymphocytes, and eosinophils, infiltrate the damaged area. Lymphocytes generate cytokines and chemokines that activate macrophages (Tanaka and Miyajima, 2016). Activated macrophages stimulate inflammatory cells to secrete pro-inflammatory cytokines and excessively activate inflammatory signals to generate ROS. In fibrosis, macrophages generate fibrosis-promoting factors such as TGF-β and PDGF, and they are found near the myofibroblasts that produce collagen (Seki and Schwabe, 2015; Koyama and Brenner, 2017). These finding suggest that macrophages are closely associated with the activation of myofibroblasts.

Macrophages exhibit a heterogeneous cell population with enormous cellular plasticity and various microenvironmental stimuli cause them to polarize into different phenotypes (Zhu et al., 2015; Viola et al., 2019). Hepatic macrophages consist of Kupffer cells (liver-resident macrophages) and circulating monocytes (inflammatory recruiting macrophages) (Beattie et al., 2016). Both cells can activate HSCs, and they can be transdifferentiated by TGF-β. Resident hepatic macrophages recruit monocytes to promote liver fibrosis by secreting the chemokine CCL2 (Cai et al., 2018; Wen et al., 2021).

Macrophages are classified into pro-inflammatory M1 type and anti-inflammatory M2 type (Chun and Kim, 2018; Yu et al., 2018). M1 macrophages are predominant during liver injury and promote EMT and ECM deposition, whereas M2 macrophages release anti-inflammatory factors including IL-10, arginase, and HO-1 (Li et al., 2017; Chen et al., 2020). However, during chronic liver injury, M2 macrophages can contribute to the production of pro-fibrotic factors, especially TGF-β and PDGF (Sun et al., 2017).

Thus, macrophages are the primary sources of TGF-β, and they are important contributors to the development of liver fibrosis (Seki and Schwabe, 2015; Koyama and Brenner, 2017). TGF-β can cause macrophage polarization towards a M2-like phenotype through SNAIL1 (Zhang et al., 2016). SNAIL overexpression in human THP-1 macrophages stimulates the expression of M2 markers (such as CD206) and anti-inflammatory IL-10. Conversely, SNAIL deficiency triggers M1 polarization by enhancing pro-inflammatory cytokines.