Zonation, Zonation, Zonation: The Real Estate of the Liver

Abstract

The liver serves as a central hub for a diverse set of functions including metabolic homeostasis, detoxification, and protein synthesis. While appearing homogeneous, hepatocytes, the major workhorse in the liver, demonstrate spatial identity within the lobule, which in turn dictates gene and protein expression and, eventually, function. Presenting as an axis from the portal triad to the central vein, this organization has been conventionally referred to as metabolic zonation. In recent years, the heterogeneity in expression and function is now understood to extend well beyond hepatocytes and metabolism to include nonparenchymal cells and diverse functions. Although the lobule is conventionally divided into three zones, spatial multi-omics technologies reveal a more nuanced picture, where zonation provides a coordinate system for an eclectic but highly functional hepatic milieu. We summarize the current understanding of liver zonation as it contributes to division of labor, injury compartmentalization, and stepwise arrangement of metabolic pathways and discuss the implications of this framework for liver homeostasis, regeneration, and disease.

INTRODUCTION

The liver is a remarkable organ that is responsible for more than 500 functions including essential homeostatic processes such as detoxification, energy metabolism, protein synthesis, bile acid metabolism, and iron homeostasis. It demonstrates a distinct spatial organization at a cellular and molecular level referred to as metabolic zonation or simply zonation. Conceptually, the liver may be divided into functional units called lobules, each of which contains a central vein surrounded by portal triads. From the portal triad to the central vein, the lobule is categorized broadly into zones 1, 2, and 3, where zone 1 encompasses a third of the region surrounding the portal triad, zone 3 circumscribes another third of the area around the central vein, and zone 2 captures the intermediate region (Figure 1a). There have been many studies spanning last two decades that have contributed to our understanding of the concept of zonation. Such studies have either identified molecular signals that are driving gene expression specific to the zones or improved our understanding of the predominant functions of each zone (1, 2). While the boundaries between zones are not precisely defined, this schema provides a framework by which we can understand the profound spatial heterogeneity of the liver. Zonation provides a number of functional benefits including division of labor, compartmentalization of injury, and stepwise segregation of metabolic pathways, acting as a “conveyor belt” (Figure 1b). It is intriguing to note that several basic metabolic functions of the liver have been ascribed to a particular zone even within the same broad category. For example. traditionally, based on gene expression studies from isolated cells or other modalities, for glucose metabolism, glycolysis is attributed to zone 3 and gluconeogenesis to zone 1. Likewise, for lipid metabolism, fatty acid oxidation is broadly thought to occur in zone 1 and lipogenesis in zone 3. Similarly, for amino acid metabolism, urea synthesis is credited to zone 1 and glutamine synthesis to zone 3. More recent studies have also begun to describe aberrant zonation as both a cause and an effect in various hepatic pathologies, underscoring the significance of a comprehensive understanding of the process of zonation (3, 4). While liver zonation has been known to exist for decades, the advent of multi-omics technologies, particularly spatial transcriptomics, has provided an unprecedented perspective into liver zonation, both validating prior discoveries and challenging others. Further, these technologies have helped identify unique expression of genes and proteins not only in hepatocytes but also in all nonparenchymal cells of the liver, especially endothelial cells, stellate cells, and macrophages (Figure 1c).

Figure 1.

Liver zonation provides division of labor, contributes to compartmentalization of injury, and serves as a “conveyor belt” for stepwise metabolic processes such as bile acid synthesis. (a) The liver is classically divided into three zones based on location within the porto-central axis. (b) Each zone engages in specialized functions (urea cycle, mitochondrial β-oxidation, and gluconeogenesis in zone 1; iron homeostasis in zone 2; and xenobiotic metabolism, glutamine synthesis, bile acid synthesis, peroxisomal β-oxidation, and glycolysis in zone 3), while also having unique injury susceptibility. Bile acid synthesis enzymes have been observed to be spatially arranged approximately in a stepwise manner, demonstrating that bile acid intermediates are sequentially passed along in the direction of the portal triad, where they are ultimately conjugated and deposited into bile ducts. (c) While zonation has been primarily studied in hepatocytes, a number of nonparenchymal populations have demonstrated zonated gene expression and function. Both HSC-derived Rspo3 and EC-derived Wnt2 and Wnt9b have been demonstrated to promote zone 3 hepatocyte gene expression via Wnt/β-catenin signaling. Abbreviations: BA, biliary atresia; CV, central vein; EC, endothelial cell; HA, hepatic artery; HSC, hepatic stellate cell; KC, Kupffer cell; MASLD, metabolic-dysfunction associated steatotic liver disease; PBC, primary biliary cholangitis; PSC, primary sclerosing cholangitis; PV, portal vein; SEC, sinusoidal endothelial cell. Figure adapted from images created in BioRender; Yasaka, T. 2025. Panels a and b: https://BioRender.com/9o6s9zl. Panel c: https://BioRender.com/qlt7z10.

Spatial transcriptomics and other spatial omics technologies have transformed our understanding of liver zonation (Figure 2a). For example, using a computational approach integrating single-molecule fluorescence in situ hybridization (smFISH) of landmark zonation genes and single-cell RNA sequencing (scRNA-seq), Halpern et al. (5) found that 50% of genes in the liver demonstrate a zonated expression pattern, and they also discovered zonated genes with a nonmonotonic pattern (i.e., peaking in expression at neither the central nor portal vein). Using a sequencing-based spatial transcriptomics approach, Hildebrandt et al. (6) further demonstrated the striking impact of zonation on gene expression throughout the lobule. In 2024, Xu et al. (7) further explored the homeostatic liver through sequencing-based spatial transcriptomics. This rich dataset confirmed the significant heterogeneity in zonated gene expression while uncovering new zonated genes and exploring the regulation of zonated genes through computational analysis of gene regulatory networks. These gene regulatory networks were characterized in greater depth by Bravo González-Blas et al. (8) by combining single-nucleus RNA sequencing (snRNA-seq) with single-nucleus assay for transposase-accessible chromatin with sequencing (snATAC-seq). In this work, the authors used computational and experimental approaches to explore the complex regulation of zonated gene expression. Temporal patterns in zonated gene expression have also been investigated, revealing a circadian rhythm interacting with, but independent from, liver zonation (9).

Figure 2.

Heterogeneous expression patterns of various zonated genes in the murine liver. (a) Multi-omics single-cell and spatial technologies can now be used to study liver zonation. (b) Gene expression heterogeneity across the lobule, as characterized by single-cell and spatial multi-omics, demonstrating that zonation is a complex phenomenon rather than a discrete set of well-defined regions (5-7). Even canonical zonation markers, such as zone 3 markers Glul and Cyp2e1, demonstrate varying spatial expression patterns across the lobule. Abbreviations: CV, central vein; m/z, mass-to-charge ratio; PT, portal triad; UMAP, uniform manifold approximation and projection. Figure adapted from images created in BioRender; Yasaka, T. 2025. https://BioRender.com/tekm8b3.

Spatial multi-omics studies have revealed complexity and nuance in liver zonation, challenging the orthodoxy of discrete zones. Indeed, these technologies have revealed that zonation exists not as separate compartments, but as a gradient that serves as a scaffold upon which a variety of zonation paradigms are constructed. Rather than taking zonation as a set of distinct regions, we propose that it serves as a coordinate system for the lobule. Using some set of signals, cells are able to approximate their relative position within the lobule, which dictates gene expression and function (Figure 2b).

INJURY COMPARTMENTALIZATION, DIVISION OF LABOR, AND CONVEYOR BELT

Xenobiotic Metabolism

As the predominant site of first-pass metabolism, the liver generally receives the highest exposure to exogenous compounds and the resultant toxicity. These substances undergo chemical modification, which is classified as either hydrolysis, reduction, oxidation, or conjugation (10). While these processes neutralize many toxins, they can also produce reactive intermediates, damaging the liver. Toxic liver injury can occur through a variety of mechanisms, including oxidative stress and biomolecule damage, ultimately resulting in cell death. Due to the expression gradient of various genes, zonation enables compartmentalization of injury in response to certain xenobiotics, allowing the cells in surviving zones to proliferate and allow regeneration. For example, cytochrome P450 (CYP) enzymes catalyze oxidative reactions in the liver, transferring electrons from xenobiotic compounds to molecular oxygen. CYPs tend to be expressed toward the central vein and act upon 90% of drugs (10). Here, CYP2E1, CYP1A2, and CYP3A4 are responsible for converting acetaminophen into N-acetyl-p-benzoquinone imine (NAPQI), the toxic intermediate responsible for acetaminophen toxicity (2, 11, 12). As a result, in sublethal doses, NAPQI production is restricted to layers of hepatocytes around the central vein, allowing the surrounding hepatocytes to maintain homeostasis while those at the necrotic interface proliferate and restore zone 3. It is only when the number of damaged hepatocytes exceeds a threshold that the liver is unable to undergo regeneration quickly enough, and liver failure ensues (13).

Ammonia Detoxification

The liver is the site of systemic ammonia detoxification through the urea cycle. Protein breakdown in the intestine yields amino acids, and deamination of amino acids produces ammonia, a highly toxic by-product that travels to the liver through the portal vein. A major function of the liver is to detoxify ammonia through entry into the urea cycle for the generation of urea and for eventual excretion into the urine. Urea cycle enzymes such as arginase, argininosuccinate synthase 1, and the rate-limiting enzyme carbamoyl-phosphate synthase 1 are expressed in hepatocytes closest to the ingress around the portal vein (5, 14). Additionally, glutaminase (GLS), which converts glutamine from peripheral tissues into glutamate and ammonia, is also present here, so the ammonia generated can enter the urea cycle locally to produce urea for excretion (15). As a fail-safe, glutamine synthetase (GS) is strongly expressed and limited to one to two hepatocyte layers around the central vein in zone 3. GS condenses a molecule of ammonia to glutamate to generate glutamine, thereby preventing ammonia from entering systemic circulation. Overall, this division of labor allows the portal zone to package waste nitrogen into urea while the central zone scavenges any remaining ammonia into glutamine, preventing its systemic leakage (14-16).

Bile Acid Synthesis

While blood flows from the portal triad to the central vein in the sinusoids, bile flows in the opposite direction in the biliary canaliculi along the apical surface of the hepatocytes. Bile eventually exits the liver through the hierarchical small-order to larger-order bile ducts to be deposited in the gallbladder. Early studies suggested that bile acid synthesis occurred in zone 1 hepatocytes (16); however, several studies have since consistently demonstrated a gradient of expression of bile acid synthesis and metabolism enzymes Cyp7a1, Cyp27a1, Cyp8b1, and Cyp7b1 from zone 3 to zone 1 (5, 7, 17-19). The sequence of enzymes suggests that pericentral hepatocytes synthesize bile acids, which then flow toward the bile duct in zone 1, passing along sequential intermediates in a “conveyor belt” fashion (5).

Lipid Metabolism

Fatty acid oxidation or β-oxidation is the metabolic process by which fatty acids are broken down in the mitochondria (or alternatively, peroxisomes) to produce ATP through oxidative phosphorylation and generate acetyl coenzyme A, which then enters the citric acid cycle. β-oxidation is tightly regulated and plays a crucial role in energy metabolism, especially during fasting or prolonged exercise when carbohydrates are scarce (20, 21). In the liver, β-oxidation is believed to be concentrated in zone 1, where the oxygen supply is most abundant (2). This is based on a series of experiments performed in rats using digitonin-based perfusion or microdissection to isolate periportal and pericentral hepatocytes (22). While several of these studies did report a periportal preponderance, others were not able to reproduce these results (22). Inconsistencies were also observed between male and female rats, as well as between fed, fasted, and refed states (22). Modern omics technologies have enabled a renewed investigation into the localization of β-oxidation within the lobule. In 2021, while using scRNA-seq to map the architecture of the liver through time and space, Droin et al. (9) observed that the KEGG “fatty acid degradation” gene set demonstrated a strong pericentral bias, along with the majority of lipid-related gene sets. These findings were corroborated in the spatial transcriptomics study conducted by Xu et al. (7) in 2024, where the majority of the 38 genes in this signature were heavily biased toward zone 3. The expression of Cpt1a (carnitine palmitoyltransferase 1A), encoding the rate-limiting enzyme, was highest around the portal vein and is in accordance with the aforementioned studies, several of which used CPT1 activity as a readout of β-oxidation. The orientation of fatty acid metabolism toward zone 3 was evident at the protein level using single-cell spatial mass spectrometry to investigate the zonation of the liver proteome (23). The authors attributed this finding to peroxisomal β-oxidation, in line with the previous finding in rats showing that peroxisomal oxidation consistently occurs preferentially in zone 3 (24).

De novo lipogenesis (DNL) refers to the process by which fatty acids are synthesized from carbohydrates. DNL occurs primarily in the liver and adipose tissue, allowing excess carbohydrates to be converted to fatty acids and packaged in triglycerides for energy storage (25). Similar to β-oxidation, the zonation of DNL genes has been difficult to precisely characterize. The initial studies performed in rats led to conclusions of a pericentral predominance, but other studies reported higher enzyme activities in the periportal zone (22). Gene set enrichment of the “biosynthesis of unsaturated fatty acids” KEGG pathway in the Halpern et al. (5) study showed a pericentral bias, but it had a nonmonotonic expression pattern. Droin et al. (9) also found this pathway to be enriched in pericentral hepatocytes. Massalha et al. (26) found that the genes sterol regulatory element-binding transcription factor 1 (SREBF1), ATP citrate lyase (ACLY), fatty acid synthase (FASN), and acetyl-CoA carboxylase 1 (ACACA), encoding key DNL enzymes, were pericentral in expression in humans but periportal in mice, which was also verified in mice by Xu et al. (7). This study (7) also reported a zone 2 preponderance of the “fatty acid biosynthesis” KEGG pathway. Kang et al. (27) also reported a higher abundance of mRNAs and proteins involved in DNL in zone 1 hepatocytes of mice but observed a higher level of serine 79 phosphorylation of ACC1, an inhibitory modification that suppresses lipogenesis. These findings support lipogenesis to be preferentially occurring in zone 3 hepatocytes. Berndt et al. (28) utilized shotgun proteomics to identify hundreds of differentially expressed enzymes and identified that both zone 1 and zone 3 hepatocytes are capable of free fatty acid uptake, although zone 3 hepatocytes become saturated at a high external free fatty acid level, thus contradicting higher fatty acid uptake in zone 3. Additionally, zone 1 hepatocytes were noted to possess a higher fraction of synthesized triacylglycerol, which also contradicts previous publications reporting that DNL is more pronounced in zone 3 hepatocytes (28).

Altogether, these studies suggest that while mitochondrial β-oxidation may be concentrated in zone 1, peroxisomal β-oxidation predominates in zone 3. The zonation of DNL remains elusive and may demonstrate a more nuanced pattern depending on a variety of factors including species, fed/fasted state, and sex. The relative contributions of lipid metabolic process localizations in homeostasis and disease warrant further investigation.

Glycolysis and Gluconeogenesis

One of the major functions of the liver is to maintain glucose homeostasis. Other than glycogen storage and breakdown, breakdown of glucose to produce pyruvate by glycolysis and production of glucose from noncarbohydrate sources are major functions of hepatocytes. Based on expression of key genes, gluconeogenesis, which is essential to maintain glucose levels especially during fasting or periods of intense activity, has been ascribed to zone 1. These include higher activities of key enzymes including glucose-6-phosphatase and fructose-1,6-bisphosphatase, as well as upstream enzymes such as lactate dehydrogenase and alanine aminotransferase (29). More recent spatial transcriptomic studies validate these findings (7).

Glycolysis uses glucose to produce pyruvate or lactate for energy production and has been suggested to predominate in zone 3. Jungermann & Katz (29) inferred this primarily from the higher activities of glucokinase and increased conversion of glucose to pyruvate in zone 3. However, several of the key enzymes did not show a clear zonation pattern. Further, spatial transcriptomic studies have not reported glycolysis gene sets to be enriched in any particular zone (5, 7). Our own independent analysis of the published supplemental data in one study (7) also failed to identify a strong zonation trend for the HALLMARK_GLYCOLYSIS gene set. Therefore, while the existing evidence does suggest higher levels of glycolysis in zone 3, the data to support this conclusion are currently limited, requiring an in-depth investigation.

Iron Homeostasis

Iron plays a crucial role in health and disease, and its levels are tightly regulated. Iron is absorbed from the diet through the duodenum and stored in the liver. Macrophages facilitate recycling of iron during erythrocyte turnover (30). Hepcidin has been identified as the master regulator of iron metabolism (31). It is primarily expressed in the liver, where it is secreted and enters systemic circulation. Hepcidin binds to the transmembrane protein ferroportin, which is responsible for exporting iron out of cells. In the mouse liver, hepcidin was one of the nonmonotonic genes identified by Halpern et al. (5) (i.e., peaking in expression at neither the portal nor central veins). Subsequent papers have confirmed hepcidin as a midzonal marker at the RNA level (7, 32). Thus, the regulation of iron metabolism may be one of the central roles that zone 2 serves during homeostasis.

REGULATION OF LIVER ZONATION

Wnt/β-Catenin

The canonical Wnt/β-catenin pathway is characterized by autocrine and paracrine Wnt signaling, which results in the nuclear translocation of β-catenin and activation of target genes. In the absence of Wnt ligands, cytoplasmic β-catenin is phosphorylated by the β-catenin destruction complex—consisting of AXIN1, GSK3β, CK1α, and APC—targeting it for ubiquitination. However, when frizzled (FZD) receptor and low-density lipoprotein receptor-related protein 5 (LRP5) or LRP6 coreceptors are bound by an extracellular Wnt ligand, the degradation complex is recruited to the cell membrane and inactivated. β-Catenin accumulates in the cytoplasm and translocates to the nucleus, where it serves as a cofactor for TCF/LEF (T cell factor/lymphoid enhancer factor) family of transcription factors to induce target gene expression (33).

β-Catenin was first considered to be a regulator of metabolic zonation when three groups independently observed loss of zone 3 targets when β-catenin was conditionally deleted or gain of zone 3 targets when APC was conditionally deleted from hepatocytes (14, 34, 35). Specifically, pericentral markers GS, CYP2E1, and CYP1A2 were lost in β-catenin knockout mice (34). Conversely, liver-specific knockout of APC led to β-catenin stabilization and led to expansion of pericentral genes (e.g., Glul, Axin2) across the lobule, while periportal genes (e.g., Arg1, Cps1) were scarcely detectable (14).

Neither any of the 10 Fzds nor Lrp5/6 show any zone-3 preferential expression on hepatocytes (36). However, Yang et al. (37) demonstrated that concurrent loss of hepatocytes Lrp5 and Lrp6 resulted in abrogation of zone-3 markers, while endothelial cell-specific knockout of Wls (a chaperone for Wnt secretion) using Tie2-Cre^+/−/Wls^flox/flox mice was embryonic lethal. This was followed up by several studies in which Wls was deleted from hepatic endothelial cells. These mice presented with absent zone-3 markers including GS and CYP2E1 (38, 39). Wang et al. (40) identified Wnt2 and Wnt9b in the perivenous endothelial cells. Hu et al. (19) then confirmed that simultaneous deletion of endothelial cells Wnt2 and Wnt9b suppressed zone-3 markers while promoting ectopic expression of zone-1 markers in zone-3 hepatocytes. Together, these studies elucidate a key pathway regulating metabolic zonation, whereby WNT2 and WNT9b from endothelial cells lining the central vein as well as the sinusoidal endothelial cells (SECs) in zone 3 bind to FZD and LRP5/6 coreceptors on neighboring hepatocytes, promoting expression of β-catenin targets.

β-Catenin influences gene expression through interaction with the TCF/LEF family of transcription factors (33). Gougelet et al. (41) performed chromatin immunoprecipitation combined with DNA sequencing in mice lacking either β-catenin or APC to investigate TCF4 binding patterns. They found that TCF4 interacts with hepatocyte nuclear factor 4 alpha (HNF4α) to promote expression of periportal genes at HNF4-responsive element sites, while interaction with β-catenin promoted expression of pericentral genes at Wnt-responsive element sites. Bravo González-Blas et al. (8), based on snRNA-seq and snATAC-seq data, identified Tcf7l2 (encoding TCF4) as a major regulon in liver zonation. They further identified Tcf7l1 (encoding TCF3) as another major regulon, which appears to function by repression of target genes, as evidenced by the inverse relationship between its expression (concentrated in zone 1) and the expression of its target genes (concentrated in zone 3). Intriguingly, one of the few differences between TCF3 and TCF4 is that TCF4 includes a β-catenin binding site (42). This may help to explain why Tcf7l2 (TCF4), but not Tcf7l1 (TCF3), is expressed in zone 3 (8). Of additional interest, Tbx3 emerged as the counterpart to Tcf7l1, with its expression concentrated in zone 3 while its transcriptional targets are expressed in zone 1. This result, too, is consistent with existing studies. Liang et al. (43) showed that Tbx3 is upregulated in β-catenin-mutated hepatocellular carcinoma, functioning as a tumor suppressor via inhibition of Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ), while Suzuki et al. (44) found that deletion of Tbx3 resulted in runaway differentiation of hepatoblasts into cholangiocytes. Xu et al. (7) have additionally identified potential midzone regulons, including Tbl1xr1, Atf2, and Cebpa. Much of these observations will need further functional validation through gain- or loss-of-function studies.

R-spondins

R-spondins are a class of secreted proteins that function as agonists of Wnt/β-catenin signaling (45). In 2015, Rocha et al. (46) found that R-spondin 3 (Rspo3) is expressed in endothelial cells around the central vein, and loss of Rspo3 results in an expanded zone 1 and diminished zone 3 using multiple zonation markers. Similarly, they found that ectopic Rspo1 expression resulted in zone 3 expansion. This work was followed up by Planas-Paz et al. (47), who demonstrated that R-spondins promote Wnt/β-catenin signaling in the liver through the RSPO-LGR4/5-ZNRF3/RNF43 module. More recently, the regulatory function of hepatic stellate cells (HSCs) in maintaining hepatocyte zonation was reported. Sugimoto et al. (48) demonstrated that HSCs regulate hepatic Wnt activity and zonation through a pericentral-to-periportal gradient of Rspo3 in HSCs, along with enriched Lgr4 and Lgr5 expression in pericentral hepatocytes. As a result, HSC-specific knockout of Rspo3 led to decreased expression of zone-3 genes while expanding zone-1 markers. The zonal gradient of Rspo3 in HSCs was found to be regulated by the periportal-to-pericentral gradient of transforming growth factor beta (TGFβ), which suppressed Rspo3 expression in zone-1 HSCs. This study highlights that pericentral HSC-derived Rspo3 also contributes to zonation.

Hedgehog

Like Wnt signaling, hedgehog signaling is also key in cell differentiation, proliferation, and survival. The three ligands in this family include the sonic hedgehog (SHH), Indian hedgehog (IHH), and desert hedgehog. They are similarly produced, secreted, and processed, and they act by binding to patched (PTCH1/2) and smoothened (SMO), following which the transcription factor GLI2/3 is activated. While these ligands can have redundant functions, they can also have distinct functions due to their unique tissue-specific distribution and target genes (49). There is some evidence to suggest that the SHH signaling pathway also plays a role in maintaining the gradient of liver zonation. The role of the pathway was first established in liver development, regeneration, and cancer; it was believed to be minimally active during adult homeostasis (50) but was implicated in the development of fibrosis and steatosis (51-53). Matz-Soja et al. (54) showed SHH signaling to be concentrated around the portal vein and that SHH and Wnt/β-catenin negatively regulated each other to maintain zonation. In a follow-up study, they showed that IHH is both concentrated around the central vein and broadens its region of expression in a β-catenin overactivation model (55). However, SHH ligand, PTCH1, PTCH2, SMO, and GLI3 were found to have a zone-1 preponderance (56). Using a liver-specific Smo knockout model, the authors found some evidence of increased Wnt/β-catenin signaling in response to disrupted hedgehog signaling. A mathematical model based on a series of small interfering RNA–mediated knockdown experiments of hedgehog and Wnt/β-catenin targets in isolated mouse hepatocytes suggested a paradigm in which Wnt activation promoted zone-3 expansion, while hedgehog inhibition promoted zone-1 expansion (56), which merits further exploration.

Hippo

The hepatocytes surrounding the portal triad have a unique developmental origin. They arise from the layer of hepatoblasts in the ductal plate that were positioned just far enough from the portal mesenchyme to not receive sufficient inductive signals to differentiate into cholangiocytes. These hepatocytes have been shown to express cholangiocyte markers such as Sox9 (57) and YAP of the Hippo pathway (58). Conditional deletion of Yap from hepatocytes led to expansion of Wnt/β-catenin target GS (58). Another study found that loss of hepatic CAPZ, an F-actin capping protein, resulted in YAP activation and decreased expression of pericentral markers, while YAP deletion rescued this phenotype (59). These findings have been contested, as others have reported no zonation changes in models of simultaneous YAP and TAZ deletions (60). Additional studies are thus warranted.

Oxygen Gradient

Greater than 70% of the blood supply to the liver is from the portal vein, while the rest is from the hepatic artery. This unique dual supply establishes an interesting oxygen gradient within the hepatic lobule. As the blood from the portal vein and hepatic artery traverses the sinusoids on its path to the central vein, there is almost a 30 mm Hg drop in pO₂ from the portal triad to the central vein (1). Indeed, zone-3 liver remains susceptible to hypoxic insults and ischemic injuries (1). While a central role of this oxygen gradient in regulating liver zonation has long been speculated, its precise function in regulating zonation remains an area of ongoing investigation. Hepatocytes appear to have adapted to this oxygen gradient, as zone-1 hepatocytes have a higher mitochondrial oxygen consumption rate and higher maximal respiration compared with hepatocytes in zone 3 (28). On the other hand, higher triglycerides in zone-3 hepatocytes with larger lipid drops could be a result of decreased lipid oxidation. Hypoxia inducible factors (HIFs) have been shown to display a gradient and to regulate zonated metabolic processes including gluconeogenesis and lipogenesis. Further, hypoxia has been shown to activate β-catenin signaling in hepatocellular carcinoma (61, 62). Some have hypothesized a mechanism by which hypoxia-induced HIF activity promotes repression of APC, resulting in β-catenin activation (1). Nonetheless, whether oxygen, or lack thereof, directly regulates metabolic zonation and whether this regulation involves Wnt/β-catenin signaling remain open questions.

Miscellaneous

The portal vein also is the source of hormones from the pancreas and gut. As a result, hormone gradients across the lobule have been suggested to also regulate zonation. In 2018, Cheng et al. (63) showed that in glucagon-deficient mice, zone 3 expanded at the expense of zone 1. Further, they found that infusion of glucagon in these mice could reverse these changes. Further studies are warranted to clarify whether glucagon acts directly on hepatocytes to induce these effects or whether nonparenchymal cells such as endothelial cells serve as intermediate effectors. Whether insulin and growth factors such as epidermal growth factor or hepatocyte growth factor and others, or a balance of all of the aforementioned factors, may in fact dictate the patterning of zonation needs further and systematic evaluation.

ZONATION AND REGENERATION

Zonation of Proliferating Cells

The liver has a remarkable ability to regenerate, even after losing up to two-thirds of its original mass (64). Liver regeneration (LR) can occur in response to a variety of insults, including surgical and drug-induced liver injury. LR allows resection of localized cancer via partial hepatectomy (PH) as well as transplantation of healthy donor liver (living donor or split liver) to a recipient (65). However, in chronic injury, in the presence of progressive inflammation and fibrosis, chronic regeneration can lead to the development of hepatocellular carcinoma (HCC) (66).

Since the 1970s, LR has been understood to be a zonated process. In rats, following PH a “wave” of hepatocyte proliferation is described to occur from zone 1 to zone 3 (67). More recent studies in mice have investigated these zonated kinetics in greater depth, demonstrating that proliferation occurs predominantly in zones 1 and 2 at 40 h post-PH, occurring panzonally by 72 h (68, 69). Using a fate-tracing model, Lin et al. (70) have shown that the cell population labeled by expression of Igfbp2, a zone-2 marker, expands following PH, suggesting that zone 2 contributes preferentially to LR.

During homeostasis, minimal hepatocyte proliferation is sufficient to replace lost hepatocytes. One study reported that pericentral Axin2⁺ stem cells were responsible for homeostatic repopulation (40); however, these results were disputed by others who demonstrated that homeostatic regeneration can be performed by hepatocytes across the entire lobule (71). Another study found that midlobular Hamp2+ hepatocyte populations expanded modestly over a 52-week period, while Gls2⁺ (zone 1) and Glul⁺ (zone 3) hepatocyte populations decreased, suggesting that zone 2 may be contributing marginally more to homeostatic hepatic turnover (32). These results were corroborated by others as well, who found that while all zones contribute to homeostatic regeneration, zone 2 is marginally more proliferative (68).

In the case of zone-1 injury using a 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) chronic injury model, it was reported that cell populations labeled by Gls2 expression contracted, while those labeled with Hamp2 and Glul expanded (32). Preferential zone-2 contribution following DDC injury was further demonstrated in Igfbp2-labeled cells (70). Conversely, in the case of zone-3 injury using a CCl₄ injury model, Sox9-labeled periportal hepatocyte populations, which were described as “hybrid hepatocytes,” expanded (57). Contraction of Glul-labeled populations in the CCl₄ model along with the expansion of Hamp2- and Gls2-labeled populations was shown (32). Finally, an Igfbp2-labeled hepatocyte population was shown to expand in both CCl₄ and acetaminophen injury models (70).

Overall, these results suggest that while all zones have regenerative capacity, zone 2 appears to be the most proliferative. After any given injury, the remaining cells will proliferate to restore the liver mass, with higher contribution from zone 2. Indeed, midzonal cyclin D1 in a normal baseline liver has been reported by us and others as well and could be contributing to cells primed to enter the cell cycle for both homeostasis and repair in this relatively protected area between zones 1 and 3 of the hepatic lobule (39, 69, 72, 73). However, in a zone-specific injury, the zones that are spared will contribute disproportionately to regeneration. This model is summarized in Figure 3a. The higher regenerative capacity of zone 2 may reflect the fact that the majority of liver injuries are initially concentrated in either zone 1 (e.g., cholangiopathies and hepatitis) or zone 3 [hypoxia, drug-induced liver injury, and metabolic dysfunction-associated steatotic liver disease (MASLD)] (3, 74-76). Thus, a primary function of zone 2 may be to provide a reserve of hepatocytes for regeneration.

Figure 3.

Liver zonation and regeneration. (a) Zone-specific contributions to liver regeneration in homeostatic turnover and injury. All zones have regenerative capacity and will proliferate in response to loss of liver mass, but zone 2 contributes disproportionately. While the relative contribution of zones to regeneration may be imbalanced, overall zonation is maintained in response to acute injury.

(b) Zonated regeneration kinetics following partial hepatectomy, as measured by cyclin D1 and Ki-67, in WT and hepatocyte β-catenin KO mice. Loss of β-catenin disrupts these kinetics both temporally and spatially. Abbreviations: CV, central vein; KO, knockout; PV, portal vein; WT, wild type. Figure adapted from images created in BioRender; Yasaka, T. 2025. https://BioRender.com/e19c2mg.

Control of Regeneration

Cyclin D1, a β-catenin target expressed normally in zone 2 at baseline, is important in G1 to S phase transition in hepatocytes and hence critical in LR at 12–24 h after PH. Following acute loss of β-catenin from hepatocytes, there is a defect in cyclin-D1 upregulation and in LR (Figure 3b). In wild-type mice, Wnt2 and Wnt9b both increase in expression in endothelial cells and in macrophages, especially at 12 h post-PH (19, 69). This activates β-catenin in hepatocytes and leads to increased expression of cyclin D1 from the midzone toward the portal vein at 24–40 h and panzonal expression at approximately 72–96 h. BrdU (5-bromo-2′-deoxyuridine) labeling during LR for 14 days showed all hepatocytes labeling positively in all zones, which also coincided with restoration of baseline liver-weight-to-body-weight ratio (69). In hepatocyte-specific β-catenin knockout mice, delayed cyclin-D1 upregulation was evident at 40 h after PH. Hepatocyte proliferation not only was delayed, beginning at 72–96 h, but also was limited primarily to zones 1 and 2. In these mice, continuous BrdU treatment during LR for 14 days led to almost 100% labeling of zone-1 hepatocytes, approximately 50% of zone-2 hepatocytes, and less than 10% of zone-3 hepatocytes (69). As a result, mice lacking hepatocyte β-catenin failed to return to baseline liver-weight-to-body-weight ratio by 14 days, lagging significantly behind controls. A similar delay in LR kinetics was evident in mice lacking Lrp5/6 in hepatocytes and in mice lacking endothelial cell Wls or Wnt2/9b (7, 39). These studies demonstrate an important overall role of the Wnt/β-catenin pathway in panzonal hepatocyte proliferation. While zone-1 (and to a lesser extent zone-2) hepatocytes can still proliferate through a compensatory, insulin/mTOR-dependent mechanism, zone-3 hepatocytes appear to be exclusively under the control of Wnt signaling for proliferation (69). Further studies are needed to understand the underlying mechanisms of this zone-specific Wnt/β-catenin dependence.

SHH signaling, relevant to zonation, was also shown to be required for LR through the use of pharmacologic inhibitors (77). Spatial transcriptomic analysis of cell–cell signaling in regenerating mouse livers showed upregulation of hedgehog signaling through IHH at 48 h post-PH, which was panzonal, but with a strong zone-3 preponderance. This is consistent with prior work showing IHH to be expressed pericentrally (55). Hedgehog signaling thus appears to be important in LR, but its precise role and its interactions with Wnt remain open questions.

Therapeutic Opportunities in Regeneration

The molecular mechanisms underlying regeneration are numerous and context-dependent, but there appear to be some common developmental pathways, including Wnt and hedgehog, that may be broadly proregenerative. These pathways may present therapeutic opportunities for modulation of regeneration through the use of novel agonists. For example, use of a novel tetravalent tailored antibody, FL6.13, that directly binds to frizzled receptors and LRP5/6 to activate β-catenin, rescued liver zonation, cyclin-D1 expression, and hepatocyte proliferation after PH in mice lacking endothelial cell Wls or Wnt2/9b. The same agent improved LR following acute acetaminophen overdose in mice, as indicated by decreased hepatic necrosis and improved serum chemistries by activating β-catenin, cyclin D1, and hepatocyte proliferation (19).

ZONATION OF NONPARENCHYMAL CELLS

Nonparenchymal cells (NPCs) of the liver account for approximately 10% of liver mass and play distinct roles during liver homeostasis and disease. NPCs include SECs, HSCs, and Kupffer cells (KCs), as well as other cell types. In the past, these NPCs were thought to be a homogenous population, largely due to the limited markers available. Recently, application of spatial transcriptomics combined with multi-omics studies significantly expanded the current understanding of NPC heterogeneity (Figure 1; Table 1) and their unique functions in disease progression, particularly in localized injuries, and their influences on liver zonation (78). Understanding the zonation of NPCs will provide important insights in disease pathophysiology.

Table 1.

Highlighted multi-omics studies of zonation in human and mouse liver

Species	Focus	Technology	Reference	Year
Mouse	Homeostasis	scRNA-seq, smFISH	Halpern et al. (5)	2017
Mouse	ECs in homeostasis	Paired-cell sequencing	Halpern et al. (99)	2018
Mouse	HSCs in homeostasis and fibrosis	10x Chromium	Dobie et al. (81)	2019
Mouse	ECs in homeostasis and cirrhosis	10x Chromium	Su et al. (101)	2021
Mouse	Homeostasis	High-definition spatial transcriptomics	Hildebrandt et al. (6)	2021
Mouse	Circadian rhythm	scRNA-seq, smFISH	Droin et al. (9)	2021
Mouse	ECs in homeostasis and regeneration	Molecular cartography	Hu et al. (19)	2022
Mouse	Macrophages in homeostasis and MASH/MASLD	10x Visium, molecular cartography	Guilliams et al. (111)	2022
Mouse	Aging	10x Visium	Nikopoulou et al. (174)	2023
Mouse	Gene regulatory networks	snRNA-seq, snATAC-seq	Bravo González-Blas et al. (8)	2024
Mouse	Homeostasis and regeneration	Stereo-seq	Xu et al. (7)	2024
Mouse	Lipid metabolism in MASLD	Imaging mass spectrometry	Seubnooch et al. (129)	2024
Human	Immune cells in homeostasis and PSC	10x Visium, NanoString GeoMx DSP	Andrews et al. (148)	2024
Mouse	HSC-derived Rspo3	Molecular cartography	Sugimoto et al. (48)	2025
Human	Homeostasis and fibrosis	MERFISH	Watson et al. (175)	2025

Abbreviations: DSP, digital spatial profiler; EC, endothelial cell; HSC, hepatic stellate cell; MASH, metabolic-dysfunction associated steatohepatitis; MASLD, metabolic-dysfunction associated steatotic liver disease; MERFISH, multiplexed error-robust fluorescence in situ hybridization; PSC, primary sclerosing cholangitis; Rspo3, R-spondin 3; scRNA-seq, single-cell RNA sequencing; smFISH, single-molecule fluorescence in situ hybridization; snATAC-seq, single-nucleus assay for transposase-accessible chromatin with sequencing; Stereo-seq, spatial enhanced resolution omics sequencing.

Hepatic Stellate Cells

HSCs are resident mesenchymal cells that reside in the perisinusoidal area between hepatocytes and sinusoidal endothelial cells called the space of Disse. HSCs were once considered a homogenous population that are quiescent at baseline and store vitamin K (79). Upon activation by injury, HSCs transdifferentiate into myofibroblasts and become the key source of pathologic extracellular matrix (ECM) deposition, promoting liver fibrosis and cirrhosis (80).

ScRNA-seq, snRNA-seq, and spatial transcriptomics studies have revealed heterogeneous subpopulations of quiescent and activated HSCs with unique gene signatures that may serve distinct functions at baseline and in response to injuries. Dobie et al. (81) identified two distinct subpopulations of quiescent HSCs based on zonation and exclusive gene expression profiles. Ngfr⁺ and Itgb3⁺ HSCs were localized in the periportal region and were termed portal vein–associated HSCs (PaHSCs). In contrast, Adamtsl2⁺ and Rspo3⁺ HSCs were found in the pericentral region and thus termed central vein–associated HSCs (CaHSCs). The heterogeneous subpopulations of quiescent HSCs based on zonation was further confirmed in another study using scRNA-seq that identified two populations that expressed either CaHSC markers or PaHSC markers (82). Based on the molecular interactions analysis in this study, PaHSCs had a strong interaction at the level of basement-membrane-type ECM proteins, whereas CaHSCs had strong interactions of the activator protein-1 (AP1) protein, implicating potential functional distinction between two populations. In addition, a long-term lineage tracing study using SMMHC-CreER^T2, which labeled a subset of NGFR⁺ HSCs in zone 1, revealed that these cells remained confined to zone 1 and do not migrate to zone 3 even after 1 year of labeling (83).

CCl₄-induced centrilobular fibrotic liver injury activated CaHSCs to collagen-producing myofibroblasts (81). In agreement, Tcf21⁺ pericentral HSCs served as the primary source of myofibroblasts in CCl₄-induced liver injury (84). Portal HSCs responded differently to CCl₄ treatment. Instead of differentiating into myofibroblasts, a subset of NGFR⁺ HSCs in zone 1, labeled by SMMHC-CreERT2, became proliferative and migrated throughout the lobule to localize near endothelial cells undergoing capillarization (83). Acetaminophen-induced zone-3 injury stimulated activation of HSCs in zone 2 and zone 3, while proliferation genes were upregulated in zone-1 and zone-2 HSCs (85).

The origin of myofibroblasts in bile duct ligation–induced cholestatic injury remains uncertain. While some studies suggest that portal fibroblasts are the primary source of fibrogenic myofibroblasts (86, 87), others indicate that periportal HSCs, marked by Tcf21-CreER, are the predominant contributors (84). Regardless, these findings highlight that injury responses are highly dependent on both the injury type and its location within the hepatic lobule.

In the Western diet (WD)-fed metabolic dysfunction-associated steatohepatitis (MASH) mouse model, both quiescent CaHSCs and PaHSCs were found to contribute equally to the fibrogenic, activated HSCs (82). Although the WD-induced MASH model has been shown to cause macrovesicular steatosis in zones 1 and 2 and microvesicular and macrovesicular steatosis in zone-3 hepatocytes (88, 89), the overall diet-induced injury appears to be global rather than confined to specific zones. This likely results in uniform activation of HSCs throughout the liver.

The regulatory function of HSCs in maintaining hepatocyte liver zonation has been proposed. Additionally, the role of angiocrine Wnts and R-spondins in the regulation of zone 3 has been discussed above. Zhao et al. (90) showed that Gdf2/Bmp10 expression in HSCs regulates Wnt2 and Rspo3 expression in endothelial cells, thereby regulating zone-3 metabolic zonation. Additionally, it has been demonstrated that HSCs are the major source of Rspo3 in zone 3, as discussed earlier in this article and also published elsewhere (48).

Liver Sinusoidal Endothelial Cells

The SECs lining the sinusoids are the most abundant NPCs in the liver, constituting 15–20% of all liver cells. These cells play essential roles in clearance, flow regulation, immune surveillance, and angiocrine signaling, thereby contributing to liver homeostasis, metabolism, and immunity (91, 92). Upon injury, SECs undergo functional and phenotypic changes, including capillarization, which promotes fibrosis, inflammation, and vasoconstriction (93). In the context of zonation, SECs primarily function by secreting key factors such as WNT2, WNT9b, RSPO3, and BMP6, which regulate metabolic and functional zonation in hepatocytes (47, 94-96).

Multi-omics studies have revealed the spatial heterogeneity of SECs, showing that 35% of SEC genes in healthy mice and 67% in healthy human livers exhibit significant zonation (97-99). These studies confirmed the gradient expression patterns of WNTs and RSPO3. Additionally, at baseline, SECs in zone 1 are enriched in genes involved in angiogenesis and amino acid metabolic processes, whereas SECs in zones 2 and 3 are enriched in scavenging and immune-related pathways, including innate immunity, phagocytosis, leukocyte activation, and bacterial defense (98, 100, 101).

SECs in different liver zones exhibit distinct responses to injury. SECs in zones 1, 2, and 3 of cirrhotic mice have been shown to maintain differentially expressed gene profiles (101). Zone-3 SECs had the most upregulated genes associated with capillarization and ECM, while showing decreased expression of endocytic receptors essential for scavenging functions. This suggests that zone-3 SECs may be more vulnerable to injury and inflammation. Similarly, SECs in zones 2 and 3 were shown to be more susceptible to chronic hepatitis B infection, where they upregulated genes involved in capillarization and nuclear factor kappa B (NF-κB) signaling (100). Additionally, zone-3 SECs were shown to be in close communication with natural killer (NK) and T cells via Cxcl16-Cxcl6 interactions, triggering inflammation and immune cell recruitment (100). Endothelial cells in liver tumors have also been shown to undergo transcriptomic changes to show strong similarities to vascular SECs rather than SECs losing zonated SEC gene expression (102).

The upstream regulators that control the expression of angiocrine factors in SECs have not been fully elucidated. HSCs have been shown to express GDF2 and BMP10, which interact with ALK1 receptors on SECs, subsequently promoting the expression of BMP2/6, WNT2, and RSPO3 (90). Consistently, SEC-specific deletion of Acvrl1, the gene encoding ALK1, resulted in the loss of SEC-derived angiocrine factors, including WNT2, WNT9B, and RSPO3, ultimately leading to disruption of zonation (103). In addition, an endothelial transmembrane receptor involved in cardiovascular development, Heg, was suggested as an upstream regulator of Wnt expression in SECs (104). In contrast, NOTCH signaling is suggested to play an inhibitory role on Wnt signaling. Duan et al. (105) demonstrated that c-Kit⁺ SECs facilitate LR following PH by enhanced WNT2 production. However, Notch activation in SECs resulted in impaired LR due to suppression of Wnt2 expression.

Kupffer Cells

The liver has the largest population of tissue-resident macrophages known as KCs (106, 107). KCs reside within the sinusoids and remain relatively immobile during homeostasis, primarily functioning to remove pathogens, dead cells, and toxins from the blood while also modulating immune responses to promote immune tolerance (108). Studies utilizing single-cell technologies have reported mixed findings regarding KC heterogeneity. Several studies have suggested KCs to be a heterogeneous population (97, 98, 109, 110). Using a multi-omic approach, Guilliams et al. (111) determined KCs to be relatively homogeneous and lacking zonal gene expression. Most recently, Miyamoto et al. (112) identified a spatially heterogeneous distribution of KCs with distinct functions that differ from previously identified subsets of KCs.

It is well-accepted that KCs are prominently located in zone 1, and previous emphasis has been on understanding their functional role (111-113). This asymmetric distribution was demonstrated not to be developmental but rather to arise from sustained commensal-induced MyD88-dependent signaling in SECs and chemokine gradients (113). Periportal MARCO⁺ KCs have been determined to capture bacteria entering the portal vein to prevent bacterial dissemination and suppress immune responses through production of interleukin 10 (IL-10) and sequestration of gut-derived bacterial products and metabolites (112, 113).

As resident KCs decline with age or under disease conditions, monocyte-derived macrophages (MDMs) have been shown to replenish the liver macrophage pool and become KC-like cells (106, 114). Heterogeneous populations of MDMs have been also suggested with distinct functions, although there is no clear evidence of spatial heterogeneity. MDMs are known to preferentially accumulate in zone 1 in cholestatic diseases, including primary biliary cholangitis, primary sclerosing cholangitis, and MASH/MASLD (115, 116). In MASH, lipid-associated macrophages accumulate near bile ducts and are recruited to steatotic zones in steatotic livers—pericentrally in patients and panzonally in murine models (111, 117). More recently, it was demonstrated that these macrophages were recruited to the periportal region by cholangiocytes after DDC-induced liver injury, where they enhanced the ductular reaction (118). The exact role of these cells is unclear, with some suggesting a role in pathology via acquisition of a proinflammatory phenotype (116, 119), while others suggest a protective role through enhancement of lipid metabolism in hepatocytes (117, 120). The relevance of understanding distinct, spatially oriented macrophage identities and functions remains a focus in the field.

PATHOPHYSIOLOGY

MASLD/MASH

MASLD affects more than 30% of the global population and carries a high risk for the development of cirrhosis and HCC (121). MASLD starts as simple steatosis, of which a subset progress to MASH and fibrosis (122). The pathogenesis is believed to be initiated by lipid toxicity that induces hepatocellular injury followed by compensatory hepatocyte proliferation. However, continued metabolic insult leads to chronic regeneration, inflammation, HSC activation, and progressive fibrosis (123, 124).

MASLD/MASH is associated with changes in metabolic zonation. Steatosis is often seen in zone 3 during the early stages of adult MASLD, although pediatric MASLD presents with more heterogeneous zonation patterns of steatosis (3). Mice fed a choline-deficient, l-amino acid–defined, high-fat diet (CDAA-HFD) displayed pericentralization marked by expansion of zone-3 genes and collapse of zone-1 genes in spatial transcriptomic profiling (125). Another study observed downregulation of zone-specific genes in both zone-1 and -3 hepatocytes in HFD-fed mice, with zone-3 hepatocytes specifically upregulating genes involved in lipid accumulation and the ketogenic pathway (126). It is also relevant to note that targets of therapies approved for treatment of MASLD, such as thyroid hormone receptor beta (THR-β) and peroxisome proliferator-activated receptor alpha/gamma (PPARα/γ) signaling, are also localized to zone 3 (5, 127). Lipid accumulation in zone 3 is also an initial event in adult MASLD patients (128-130). Deeper analysis to identify zonated lipid species using advanced mass spectrometry imaging revealed partial to complete loss of lipid zonation in human MASH livers. As disease progressed, diminished lipid distribution differences were evident between periportal and pericentral regions in MASH when compared with MASLD livers (128). Spatial lipidomics using murine MASH models identified significant increases in lipid accumulation with zone-specific compositions of several lipids (129). Specifically, most triacylglycerols, diacylglycerols, sphingolipids, and ceramides change their spatial distribution from zone 3 to zone 1, while phospholipids transition from zone 1 to zone 3. The exact significance of these zonal changes is unclear, but they suggest metabolic deregulation or adaptation and require further investigation.

Wnt/β-catenin signaling has been shown to drive pericentral steatosis during early MASLD. In transgenic mice with β-catenin activation, HFD-induced steatosis was enhanced in zone 3, likely due to increased expression of glycolytic and lipogenic genes (131). Zhou et al. (125) further demonstrated that pericentralization in CDAA-HFD-induced MASH mice was associated with upregulation of Wnt2, Wnt9b, and Rspo3, along with increased expression of β-catenin target genes in periportal and midzonal areas. Interestingly, overexpression or knockdown of Rspo3 in hepatocytes did not significantly affect steatosis, inflammation, or fibrosis. Hence, Wnt signaling may drive zonation changes rather than directly causing disease. Indeed, CDAA-HFD-fed Axin2 reporter mice showed a collapse of zone-3 hepatocytes, suggesting pericentral injury that may be triggering compensatory zone-3 expansion (132). While Wnt signaling may drive spatial reorganization, it also appears to play a protective role in disease progression. For example, deletion of Lgr4 and Lgr5, the Wnt signaling potentiators, resulted in increased hepatic steatosis and fibrosis (133). Human cirrhotic livers also show downregulation of Wnt target genes, such as CYP1A2, CYP2E1, and GLUL. A much deeper insight into Wnt-β-catenin signaling in controlling MASLD disease progression will be critical.

The transition from hepatic steatosis to steatohepatitis appears to be driven by portal inflammatory infiltrates, which induces fibrogenesis (134). Histological analysis of adult MASLD patients suggests a pattern of early pericentral steatosis with inflammation, followed by periportal inflammation and fibrosis in later stages (135). Hippo/YAP/TAZ signaling has been implicated in this portal inflammatory response. An increased number of YAP⁺ ductular cells correlated with the severity of inflammation and fibrosis in MASLD patient samples (136), as well as in murine models of advanced fibrosis (137). Moreover, MDMs were observed to infiltrate the periportal area early in steatohepatitis, followed by portal and periductal recruitment of other inflammatory cells (134). Further studies are needed to better understand the mechanistic connection between early pericentral steatosis and subsequent portal inflammation, which may be a critical step in disease progression.

Alcoholic Liver Disease

Alcohol-related liver disease (ALD) encompasses a broad spectrum of liver damage due to excessive alcohol consumption, including steatosis, steatohepatitis, hepatitis, and cirrhosis (138). Up to 20% of individuals with chronic alcohol consumption will progress to advanced ALD due to genetic, environmental, and epigenetic cofactors (138).

ALD is associated with perturbations in hepatic metabolism, as up to 20% of ethanol consumed is metabolized by the pericentral CYP2E1 enzyme (139). Seminal studies in rats demonstrated minimal changes in liver zonation marker distribution of glutamate dehydrogenase, lactate dehydrogenase, and alanine transaminase (ALT) following chronic alcohol feeding compared with controls (139). Chronic alcohol feeding, however, reduced the enzymatic activity of lactate dehydrogenase and ALT in periportal hepatocytes, while significantly increasing triacylglycerol accumulation in pericentral hepatocytes (139). Alcohol consumption dampens Wnt/β-catenin signaling and increases FOXO3-dependent hepatocyte apoptosis (140). Extrinsic activation of Wnt/β-catenin signaling reduces liver damage and inhibits hepatocyte apoptosis (140). Liver-specific β-catenin knockouts show worsened alcohol-induced injury in female mice due to excessive oxidative stress response and improper alcohol metabolism from reduced SOD-2, CYP2E1, and ADH expression (141). Conversely, Cao et al. (142) found that short-term alcohol treatment in mice led to decreased proliferation and increased CYP2E1 and SOD expression in hepatocytes. RNA-seq analysis revealed aberrant transcription factor and gene expression patterns of fatty acid synthesis, cholesterol metabolism, and complement signaling in hepatocytes of alcohol-fed mice compared with controls (142). These studies in mice demonstrate that Wnt/β-catenin signaling is essential for the regulation of liver zonation and ethanol metabolism to protect against liver injury and formed the basis of a clinical trial to use a Wnt agonist in alcoholic hepatitis (AH) patients.

Integration of scRNA-seq, snRNA-seq datasets, and spatial transcriptomics from patients with decompensated ALD cirrhosis revealed spatially defined distinct gene expression (143). Periportal hepatocytes expressed elevated BMP7, HAL, and HBZ; pericentral hepatocytes overexpressed TUFT1; and interzonal hepatocytes lacking periportal or pericentral markers overexpressed dihydrofolate reductase (DHFR) and fibroblast growth factor receptor 2 (FGFR2) (143). These data suggest alterations of metabolic capacities in ALD, implying adaptation. Independent analysis of AH livers in patients also displayed increased and diffuse distribution of cholangiocytes (ductular reaction), endothelial cells, macrophages, and hepatic stellate cells and reduced hepatocyte numbers compared with controls (144). A large proportion of PCNA⁺ KRT8⁺ KRT7⁺ hepatocytes showed transcriptomic signatures resembling cholangiocytes (144). Increased SOX9⁺ hepatocytes have been shown in AH patients, suggesting either hepatocyte-to-biliary transdifferentiation or representing biliary-to-hepatocyte reprogramming due to the extensive ductular reaction seen in these cases (145). However, these cells could simply represent dedifferentiation of mature hepatocytes in AH, as has been shown due to downregulation of a key liver-enriched transcription factor—hepatocyte nuclear factor-4α (146). These hepatobiliary cells in AH did show an inflammatory profile similar to reactive cholangiocytes, insinuating a role for hepatocyte dedifferentiation in immune cell recruitment. Overall, hepatocyte reprogramming and liver zonation alterations are hallmarks of ALD requiring further investigation.

Cholangiopathies

Cholangiopathies refer to a range of disorders affecting bile ducts leading to inflammation, fibrosis, and liver failure. Prototypical examples include primary sclerosing cholangitis (PSC), primary biliary cholangitis (PBC), and biliary atresia (147). PSC is a rare, progressive cholestatic liver disease associated with increased portal fibrosis, biliary metaplasia, and inflammation (147). Up to 70% of patients with PSC have coexisting inflammatory bowel disease, and this group is more likely to develop colorectal cancer and cholangiocarcinoma. PSC is suspected to have an immune basis. On the other hand, PBC is an autoimmune disorder associated with bile duct atrophy and cholestasis characterized by immune-mediated bile duct destruction. Biliary atresia is a pediatric liver disease resulting from inadequate development or destruction of bile ducts resulting in cholestasis and fibroinflammatory liver damage. A few studies have employed spatial omics analyses on explanted livers to increase our understanding of zonation in these diseases (148, 149).

While lipid-associated-like macrophages, KCs, and MHCII⁺ macrophages were found to reside near the pericentral zone, and activated macrophages within the periportal zone, the overall diversity of the macrophage population was enhanced in PSC and PBC compared with control livers (148). In PSC and PBC livers, scar-associated monocyte-like macrophages with specific gene expression (albeit with some disease-specific differences) colocalized near cholangiocytes in areas of periductal fibrosis. Also, PSC-enriched macrophage subclusters were predominantly monocyte-like macrophages, while both PBC and PSC livers were enriched with KCs expressing MARCO, CD5L, VCAM1, and MHCII components. Development of the first intrahepatic T cells atlas in PSC livers also revealed expansion of hepatic T cells with a unique naive CD4⁺ T cell population primed for differentiation to T_H17 cells (150). PSC and PBC scar regions displayed potential communication between T cells, HSCs, and periportal SECs (148). Increased cholangiocytes in portal regions of biliary atresia were accompanied by cell–cell contact with HSCs, endothelial cells, and TREM2⁺ macrophages compared with control livers (149). Further studies aiming at blocking such interactions are needed to truly demonstrate functionality of such intercellular communication.

Reprogramming of hepatocytes to cholangiocytes has been widely speculated in patients and shown directly through fate tracing in mice as a result of chronic bile duct injuries (151). PSC and PBC patient livers exhibit fewer periportal hepatocytes and an increase in interzonal hepatocytes that lack pericentral or periportal markers compared with controls (148). These interzonal hepatocytes, like in AH, had upregulation of cholangiocyte-associated genes and genes involved in immune regulation and activation, including tumor necrosis factor alpha (TNFα) signaling and interferon gamma (IFNγ) response pathway (148). These reprogrammed hepatocytes were also enriched in TGFβ signaling, which has been shown to play an important role in hepatocyte-to-cholangiocyte reprogramming (152). Spatial transcriptomic analysis of biliary atresia explant livers showed an expanded portal region and reduced central region compared with the controls (149). However, the authors noted that liver zonation is not well established in pediatric livers, with liver hematopoiesis persisting after birth (149). Increased SOX9⁺ hepatocytes have been found in other mouse models of cholestatic liver injury, including dual loss of β-catenin and γ-catenin in the hepatobiliary compartment or from hepatocytes in Mdr2-knockout mice, and may suggest a transition from hepatocyte to cholangiocyte to assist in bile duct repair (153).

Hepatocellular Cancer

A common end result of chronic hepatic pathologies is HCC. This occurs secondary to chronic hepatocyte proliferation in response to decades of hepatocyte insult by viruses, alcohol, metabolites, iron, etc., in the presence of chronic inflammatory milieu and free radicals. Mutations that allow the survival of cells in this adverse microenvironment eventually lead to hepatocyte transformation, dysplasia, and progression to HCC. Approximately 50% of HCCs display β-catenin activation due to gain-of-function somatic missense mutations in CTNNB1 (26–37%) or loss-of-function mutations in AXIN1/2 (5–10%) or APC (2–3%). Upregulation of the pericentral markers, especially GS, have been reported in β-catenin-active preclinical liver cancer models (154), hepatocellular adenomas (155), hepatocellular carcinomas (156, 157), and hepatoblastomas (158), which all show multifactorial β-catenin activation in these tumor subsets (159-164).

Various attempts have been made to classify HCC on a molecular basis. For example, Hoshida et al. (165) delineated the S1, S2, and S3 subtypes, and other classification schema, such as the more fine-grained system by Boyault et al. (166), generally align with the Hoshida subtypes. Notably, β-catenin-active HCCs comprise a subset of S3 tumors, which are considered to be more differentiated (165). A direct link between zonation and HCC subtypes was first presented in a 2017 study, which found two distinct subclasses of “nonproliferative” HCCs, likely capturing the S3 subtype (167). One subclass was strongly associated with β-catenin mutations and expressed zone-3 markers (perivenous-type), while the other lacked β-catenin mutations and expressed zone-1 markers (periportal-type). Another study further explored this concept through the lens of metabolism, identifying both a “rich metabolism” subclass corresponding to Hoshida S3 and a “glycolysis” subclass corresponding to Hoshida S1 (4, 168). The “rich metabolism” subclass was enriched for zonated metabolic processes, including “fatty acid metabolism,” “bile acid metabolism,” and “xenobiotic metabolism,” highlighting the likely role of zonation in HCC metabolism. Indeed, studies suggest that β-catenin-mutated HCCs do not demonstrate the Warburg effect, by which cancer cells preferentially engage in glycolysis even in the presence of oxygen (169-171). Instead, they appear to be dependent on fatty acid oxidation, and inhibition of this pathway halts HCC progression in mice (172). Indeed, single-cell spatial transcriptomic studies in both mice and humans have shown β-catenin-driven HCCs to uniquely show zone-3 hallmarks of bile acid metabolism, xenobiotic metabolism, and fatty acid oxidation (172, 173). All of these results demonstrate that understanding the link between liver zonation, metabolism, and HCC subtypes may illuminate novel therapeutic approaches.

CONCLUSION

In summary, liver zonation is an intricate and highly nuanced organizational system. While dividing the lobule into three zones is often a useful conceptual framework, biologically there do not appear to exist any such discrete boundaries. Rather, zonation appears to serve as a scaffold, upon which a multitude of spatially organized processes are implemented. Similar to how a city is a complex real estate and an ecosystem built on top of a common set of coordinates, the lobule can be considered to be the central metabolic real estate and ecosystem of the body, with liver zonation serving as the coordinate system. Each cell in the liver, using some combination of paracrine, endocrine, and likely autocrine signals, can infer its relative location within the lobule, and, through the transduction of signaling pathways (such as Wnt/β-catenin), can activate specific transcriptional programs for the given cell type and intralobular location. Modern multi-omics technologies have ushered in a new era of spatially informed biologic discovery, and much remains to be elucidated in liver homeostasis, regeneration, and disease in the context of zonation. Restoration of disrupted zonation, modulation of pathways that control zonation (such as Wnt/β-catenin and hedgehog), and targeting of zonated metabolic pathways offer promising therapeutic avenues for various liver pathologies. Further research is needed to understand the mechanisms controlling zonation and to translate these discoveries into effective treatments.