Current Opinion in Cell Biology Compartmentalization, Cooperation, and Communication: The 3Cs of Hepatocyte Zonation

Abstract

The unique architecture of the liver allows for spatial compartmentalization of its functions, also known as liver zonation. Unlike organelles, this compartment lacks a surrounding membrane, making traditional biochemical tools unavailable for studying liver zonation. Recent advancements in tissue imaging and single-cell technologies have provided new insights into the complexity of tissue organization, rich cellular composition, and the gradients that shape zonation. Hepatocyte gene expression profiles and metabolic programs differ based on their location. Non-parenchymal cells further support hepatocytes from different zones through local secretion of factors that instruct hepatocyte activities. Collectively, these elements form a cohesive and dynamic network of cell-cell interactions that vary across space, time, and disease states. This review will examine the cell biology of hepatocytes in vivo, presenting the latest discoveries and emerging principles that govern tissue-level and sub-cellular compartmentalization.

1. Introduction

Throughout history, the liver has captivated human interest; spanning from the ancient ritual of hepatoscopy, where the livers of sacrificed animals were scrutinized to foretell the future, to the medical doctrine of hepatocentrism, which positioned the liver at the core of the human body.^1,2. In 1666, Marcello Malpighi’s pioneering work in microscopic anatomy provided the first insights into the organization of liver parenchyma³. Although he speculated about the functional relationship between liver cells and the neighboring vascular structures, experimental tools were lacking to test his hypothesis.

Advancements in electron microscopy two centuries later revealed diverse hepatocyte subcellular organization leading to the concept of liver zonation—the spatial division of liver functions^4,5,6. Although compartmentalization through physical separation from the environment, via a membrane, enables the efficient and coordinated execution of diverse biochemical processes, non-canonical forms of compartmentalization, such as liver zonation, have not been as extensively studied mainly due to technological shortcomings.

Recent developments, especially in single-cell technologies and tissue imaging, now provide an unprecedented opportunity to investigate liver zonation, regulation, and interactions between various cells in the physiological niche. This review focuses on recent in vivo discoveries in hepatic cell biology, the principles governing metabolic compartments, and outstanding questions. The following review articles on liver zonation are recommended for further reading^{7, 8, 9}.

2. Liver architecture and functional compartmentalization

The liver is a multi-functional organ, involved in carbohydrate metabolism, protein synthesis, lipid homeostasis, detoxification, and more. How can a single organ fulfill so many, often conflicting functions?

The liver possesses a highly complex three-dimensional architecture (Fig 1A). The hepatic artery (HA) supplies it with oxygenated blood, and the portal vein (PV) delivers nutrient-rich blood, ingested toxins, and endocrine hormones. Branches of the HA and PV merge into sinusoids that enter the hepatic lobule (a microscopic hexagonal unit composed of hepatocytes) in the periphery and flow toward the central vein (CV; Fig 1B–C). In the opposite direction, pericentral hepatocytes produce and secrete bile into the bile canalicular (BC) network that merges into bile ducts in the portal regions. In addition to bile ducts, lymphatics, and nerves surround the lobule periphery.

Figure 1: Liver anatomy and spatial division of labor.

A. The liver is composed of hexagonal units called lobules. B-C. Each lobule is supplied by the hepatic artery (HA; red circle) and portal vein (PV; blue circle) which deliver oxygen and nutrients, respectively. Blood flows directionally toward the central vein (CV; blue circle in the center). Bile flows in the opposite direction and is collected in the bile duct (BD; green circle). The lobule is arbitrarily divided into three zones. C. Closeup on periportal (PP)-pericentral (PC) axis showing the three zones, with PP hepatocytes residing in the outer layers, PC (zone 3) adjacent to the CV, and mid-lobular (zone 2) in between. Gradients of glucagon and oxygen from the circulation and morphogens secreted in PC regions result in differential gene expression and spatial separation of conflicting metabolic processes. Peak expression of enzymes and transporters fluctuates during the day. Illustration created using Biorender.

This unique liver architecture is disrupted in individuals with metabolic-associated fatty liver disease (MAFLD), formerly known as non-alcoholic fatty liver disease (NAFLD). A 3D model of the human liver identified morphometric cellular and tissue parameters correlated with MAFLD progression and discovered topological defects in the BC network¹⁰. MAFLD samples displayed weaker cellular connectivity and dilated BC radius, which were more pronounced in the pericentral region. These microanatomical changes may hamper bile secretion and contribute to altered bile acid composition in MAFLD. The application of similar imaging techniques holds great potential in linking structural changes associated with pathological conditions.

Around 12 to 15 hepatocytes align along the sinusoid, and their metabolic activities are distributed asymmetrically along the portal-central axis (Fig 1C). Historically, the liver lobule is divided into three zones, with zones 1 and 3 including hepatocytes close to portal tracts (periportal; PP) and the central vein (pericentral; PC), respectively, and zone 2 residing in between (mid-lobular; Fig 1C). Importantly, the lobule and zones are not surrounded by a membrane and are difficult to delineate in a histological section. Hence, the surrounding anatomical features are used as coordinates to mark the functional compartment¹¹.

Single-cell sequencing technology has demonstrated that half of the genes expressed in hepatocytes are nonuniformly distributed across the lobule^{12, 13}. Different patterns of zonated expression reveal a spatial division of labor. For example, gluconeogenesis occurs in periportal hepatocytes, while glycolysis occurs in pericentral hepatocytes (Fig 1C). This separation provides the appropriate chemical environment and separates opposing metabolic processes. What drives these expression patterns and thus tissue-level compartmentalization?

3. Spatiotemporal regulation of liver zonation

The lobule’s distinctive anatomy and directional blood flow create physiological gradients along the PP-PC axis (Fig 1C) resulting in diverse environmental conditions in the tissue. For example, hypoxia-induced transcription factors (HIFs) are activated in oxygen-deprived areas and promote glycolysis, contributing to the PC compartmentalization of carbohydrate metabolism¹⁴.

Wnt/β-catenin signaling is another key regulator of liver zonation. Wnt ligands such as Wnt2, Wnt9B, and RSPO3 secreted by endothelial cells surrounding the central vein (Fig 1C), bind to hepatocyte receptors, translocate β-catenin, and regulate gene expression^15–17. Zonated expression of various regulators of β-catenin, such as the periportal expression of APC, negatively modulates this pathway¹⁸. Beta-catenin’s central role in organizing zonated gene expression was validated when its inhibition resulted in the expression of PP metabolic enzymes, and its activation resulted in a PC expression profile^{18, 19}.

In addition to Wnt morphogens, glucagon also regulates zonation. Glucagon-deficient mice display disrupted zonation profiles characterized by reduced PP gene expression and expanded glutamine synthase expression, indicating a shift towards a PC-like phenotype²⁰. Reintroducing glucagon in deficient mice reinstates liver zonation profiles. The secretion of glucagon in response to fasting emphasizes the liver’s remarkable ability to dynamically adjust zonation to adapt to changes in substrate availability. The interplay between glucagon and Wnt/β-catenin and other signaling pathways including Ras²¹, Hedgehog²², Yap²³, and Liver Kinase B1 (LKB1)²⁴ influences the spatial organization of hepatocyte zonation.

Liver zonation is also temporally regulated by the circadian clock, a genetically encoded timing system found in mammals. Internal and external cues constantly adjust the molecular clocks in hepatocytes, resulting in daily fluctuations in hepatic gene expression and functions. In mammals, signals originate from the suprachiasmatic nucleus (SCN) through nerve connections or diffusible molecular signals. Nutrient intake timing affects the liver impacting feeding patterns. Additionally, a substantial portion of the murine liver phosphoproteome is modulated by the circadian clock. This is accomplished through the activation of a cluster of kinases, resulting in the temporal coordination of signaling pathways²⁵. Acute injury and chronic disease disrupt the molecular clock (Fig 1C) and can also lead to liver dysfunction⁹.

Recent studies highlight sympathetic overactivity as a key driver of MAFLD pathogenesis and hepatic steatosis. Hepatic innervation was mapped using volume immunoimaging technology iDISCO+ and light sheet microscopy, revealing disorganization and degeneration of sympathetic nerves in experimental MAFLD, with similar findings in human fatty livers correlating with disease severity²⁶. Sympathetic degeneration was also observed in mice fed a high-fat diet, and in diabetic and obese genetic models²⁷. However, the involvement of hepatic innervation in organizing liver zonation was not examined.

In a recent study, changes in gene expression were surveyed in single-cell resolution over space and time²⁸. About 30% of hepatocyte genes were solely regulated by spatial factors, whereas roughly 20% were exclusively influenced by the time of day. Core clock genes, like BMAL1, showed strong temporal regulation and were uniformly expressed across the zones. Interestingly, around 7% of genes were affected by both space and time. For instance, the genes involved in fatty acid synthesis (ELOVL3) and mitophagy (BNIP3) were expressed at higher levels in pericentral cells, with ELOVL3 peaking during the fed period and BNIP3 reaching its peak during the fasting cycle. This dual regulation in spatial and temporal dimensions enables the liver to appropriately respond to metabolic demands and substrate availability during diurnal feeding and fasting cycles. Furthermore, key regulators of liver zonation, such as Wnt and hypoxia signaling, also exhibit dual regulation, with their targets peaking in expression in PC cells at the end of the fasting period.

Collectively, systemic gradients of oxygen, nutrients, and hormones together with locally produced morphogen gradients guide functional zonation over both spatial and temporal dimensions.

4. Cell-cell communication and cooperation shape liver zonation

Aside from hepatocytes, the liver is home to several non-parenchymal cells, each exhibiting spatial heterogeneity or zonation. Single-cell and spatial sequencing have revealed new cell populations, and how cell-cell interactions impact zonation. The highly fenestrated endothelial cells (ECs) line the sinusoid walls and allow for the exchange of materials between the circulation and hepatocytes (Fig 2A). ECs direct hepatocyte spatial organization, maintaining homeostasis and regeneration through the nonuniform secretion of Wnt ligands. Whereas Wnt9b is almost exclusively expressed in central vein ECs, Wnt2 displays a broader pattern of expression, being found in both central vein ECs and in sinusoidal ECs extending into zones 2 and 3^{15, 16} (Fig 2B). This yields Wnt gradients with higher concentrations pericentrally. The expression of Wnt receptors in hepatocytes is not zonated, suggesting that all hepatocytes have the potential to become pericentral, given the exposure to Wnt ligands²⁹. A recent study showed that both EC gene expression and phosphoproteome profiles were zonated. In particular, the levels of phosphorylated tyrosine in pericentral hepatocytes highlighted the Tie-Wnt signaling axis as a pivotal regulator of Wnt ligand production, secretion and regulation³⁰. In the human liver, several EC subtypes were identified, with one subgroup expanding in cirrhosis, localizing to the fibrotic niche. Under such conditions, these endothelial cells communicate with macrophages linked to scarring and mesenchymal cells through pro-fibrogenic signaling pathways, suggesting potential therapeutic targets³¹.

Figure 2: Cell-cell interactions and local gradients regulate liver zonation.

A. Closeup of the hepatic acinus, the three zones, and the spatial distribution of non-parenchymal cells. B. Local gradients formed by non-parenchymal cells provide instructions for liver zonation. Liver endothelial cells (LSECs) in PP regions sense microbiome-derived products and secrete chemokines that recruit Kupffer cells (KCs) to PP regions³². Stellate cells in the mid-lobular region secrete neurotrophin-3 (Ntf-3) to induce hepatocyte proliferation in zone 2³⁹. The morphogens (Wnt2, Wnt9, and RSPO3) secreted by LSECs in PC regions are responsible for the PC gene expression profile in PC hepatocytes^{15, 17, 29}.

Kupffer cells (KC) represent a specialized population of macrophages that eliminate bacteria, foreign particles, and aged red blood cells. KCs predominantly reside in the liver sinusoids, with a significant concentration in the portal regions (Fig 2B). This distribution results from the communication between KCs, the gut microbiome, and LSECs. Specifically, microbial-derived byproducts stimulate the secretion of chemokines from LSEC, that in turn recruit KS to portal regions. This strategic positioning of KC ensures the protection from systemic dissemination of bacteria³². Beyond the immune response, depletion of hepatic macrophages impede the process of tissue remodeling, delays cell proliferation, and hinders the restoration of metabolic zonation post-liver injury regeneration³³. These findings underscore the influential role of KCs in immune surveillance, injury repair, and hepatocyte metabolism.

Chronic liver disease has a profound impact on the composition and positioning of macrophage populations. In MAFLD, resident KCs are gradually replaced by a subset of bone marrow-derived macrophages called hepatic lipid-associated macrophages (LAMs)³⁴. LAMs exhibit a distinct transcriptional profile, featuring the expression of osteopontin, a chemokine and extracellular matrix protein, which is closely associated with MAFLD progression³⁴. Guilliams and colleagues showed that in healthy conditions, KCs are primarily located in the liver sinusoids, whereas LAMs are predominantly found in the portal vein and bile duct regions. However, in the context of obesity, this spatial distribution changes, with LAMs aggregating in zones of lipid buildup. Further investigation utilizing ligand-receptor analysis has identified multiple pericentral ligands that may have a role in regulating the positioning of KCs³⁵. Examples of macrophage population dynamics in cirrhosis, show various macrophage subgroups undergo expansion, including a distinct pro-fibrogenic population expressing TREM2+CD9+ markers³¹.

Hepatic stellate cells (HSC) in the space of Disse (Fig 2) play a multifaceted role by storing vitamin A and generating constituents of the extracellular matrix. Their involvement spans regeneration, fibrosis, and even cancer processes. Until recently, HSCs were regarded as a functionally homogenous population with the potential to undergo activation into collagen-secreting cells in liver injury. However, like other hepatic cells, HSCs display nonuniform distribution in the lobule, with distinct gene profiles associated with their spatial positioning. The central vein-associated population is the pathogenic collagen-producing cells in murine models of acute liver injury³⁶. The spatiotemporal patterns of HSC gene expression were also visualized post-injury revealing zonal dynamics of HSC activation, in a TGF-β-dependent manner, and the process of transdifferentiating into collagen-secreting myofibroblasts³⁷.

HSCs exhibit a dynamic shift between tumor-suppressing and tumor-promoting subpopulations during chronic liver disease, influencing cancer risk. An important aspect of the balance between these two HSC subtypes is their influence on hepatocytes. Cytokines and growth factors expressing HSCs (cyHSCs) are associated with the expression of hepatocyte growth factor (HGF) and exhibit tumor-suppressing effects in HCC³⁸. Whereas, myofibroblastic HSCs (myHSCs) are enriched in collagen I, contributing to tumor promotion in HCC. A dynamic shift in the myHSCs and cyHSCs population is observed during disease progression, with an increase in collagen I expression in myHSCs and a progressive decrease in HGF levels in cyHSCs. The imbalance between cyHSCs and myHSCs, along with their spatial distribution, may influence the development and progression of HCC. Mechanistically, collagen I interact with Discoidin Domain Receptor 1 (DDR1). Indeed, DDR1 expression is increased in HCC, and its activation by collagen degradation products contributes to tumor development. Conversely, inhibition of DDR1 blocks the activation of downstream tumorigenic pathways and reduces tumor growth³⁸. The restoration of the cellular balance and the mediators of communication may represent a new approach to mitigate the risk of HCC development

Despite their close association with liver disease, recent findings demonstrate the involvement of HSCs in hepatocyte turnover and homeostasis. Depletion of HSCs resulted in an unexpected reduction in liver size. Hepatocytes in zone 2 undergo division at a constant rate to maintain organ size, however, in the absence of HSCs the number of dividing cells was markedly reduced. HSCs secrete Neurotrophin-3 (NTF-3), which transmits a mitogenic signal to hepatocytes during normal physiological conditions. Thus HSCs form a mitogenic niche in the mid-lobular area (Fig 2) instructing hepatocytes to divide via paracrine secretion of the growth factor Ntf-3³⁹.

5. Organelle-level compartmentalization in vivo

Organelles compartmentalize the eukaryotic cells, creating favorable environments that promote specific reactions. These compartments are dynamic, and change their form to support function. However, elucidating the architecture of organelles within organs has been technically challenging. Cutting-edge Volumetric Electron Microscopy techniques now enable high-resolution visualization of subcellular arrangements in liver tissue⁴⁰. This was recently applied to study the complex architecture of the endoplasmic reticulum (ER). The ER is made of sheets and tubules, which respectively contribute to protein and lipid synthesis. Interestingly, the proportion of ER sheets and tubules dramatically changed in the liver of lean and obese mice⁴¹. Expression of Climp-63, a membrane protein that connects the ER to the cytoskeleton and promotes the formation of ER sheets, improved systemic metabolism in obese mice. This illustrates the dynamic nature of ER architecture in vivo and its connection to lipid metabolism⁴¹. The ER shape may affect lipid handling through communication with other organelles.

Inter-organelle communication via membrane contact sites facilitates the exchange of metabolites, signal transduction, lipid transport, and organelle membrane dynamics⁴². Recently, the role of inter-organelle communication in systemic lipid regulation was examined. Mitochondria are enveloped by curved sheets of rough endoplasmic reticulum (rER), termed “wrappER.” These specialized adhesion sites are rich in fatty acids and very low-density lipoprotein (VLDL) components, regulate VLDL synthesis, and adapt to fluctuations in lipid levels⁴³. Conversely, reduced contact sites between the ER and mitochondria, with decreased calcium exchange, were observed in the livers of mice subjected to a high-fat, high-sugar diet before the onset of insulin resistance and steatosis. This suggests that impaired ER-mitochondria communication could serve as an early event triggering insulin resistance and steatosis⁴⁴. These studies exemplify how remodeling of inter-organelle communication impacts systemic lipid and glucose homeostasis (Fig 3). Further work is required to uncover the mechanisms through which changes in organelle structure or inter-organelle communication impact homeostasis at the organismal level, and whether they can be therapeutically targeted.

Figure 3: Metabolic compartmentalization across scales in the liver.

A. The liver performs various metabolic tasks, many of which are incompatible. Therefore, hepatocytes organize their metabolism at different levels, including organelle, cellular, tissue, and organ. Organelle architecture and interactions through membrane contact sites (MCSs) allow the remodeling of the subcellular compartments. Endoplasmic Reticulum (ER; orange), mitochondria (green) lipid droplets (LDs; magenta). B. Advances in microscopy and molecular characterization of hepatocytes allow the examination of the various metabolic compartments in intact tissue. The multi-scale information provides an integrative view of how cellular activities and dynamics contribute to physiological processes and pathological development. Abbreviations: Glutamine Synthetase, GS. Illustration created using Biorender.

During fasting, the ER concurrently interacts with both mitochondria and lipid droplets (LDs), thereby facilitating lipid channeling⁴⁵. Like brown adipose cells,⁴⁶ hepatocytes have two distinct mitochondrial populations: peridroplet mitochondria (PDM) and cytosolic mitochondria (CM), each serving specific roles in lipid metabolism. Whether both mitochondrial populations are universally present in all hepatocytes, across the lobule, and how they interrelate with wrappER-mitochondrial interactions is not known. This is particularly important given periportal and pericentral mitochondria morphology and bioenergetic capacity are strikingly different which may affect their capacity to interact with other organelles⁴⁷. These differences extend to phosphoproteome profiles and nutrient-sensing responsiveness and likely play a role in the adaptation to nutrient supply.

Steatosis, or LD accumulation, is a major factor contributing to HCC but the factors involved are largely undefined. Berardi et al. demonstrated for the first time that LDs undergo autophagic degradation together with mitochondria, mediated by the mitophagy receptor Bnip3⁴⁸. This hitchhiking process, named “mitolipophagy”, offers mechanistic insight into how loss of Bnip3, also observed in HCC patients with worse prognosis, leads to lipid buildup that fuels tumorigenesis. Although further studies of organelle organization across the lobule are needed, it is evident that functional separation at this level allows for communicating metabolic states and flexibility to respond to physiological demands.

6. Lessons about compartmentalization from liver regeneration and cancer

Hepatocytes possess a remarkable regenerative capacity; however, the question of which cells divide to maintain tissue mass remains a topic of debate. Some studies demonstrated that hepatocytes regardless of lobular placement or ploidy status, can proliferate in response to chemical injury^{49 50 51 52 37}. Conversely, other studies proposed that different regions of the liver lobule vary in hepatocyte turnover, with zone 2 representing a primary source of new hepatocytes during homeostasis and regeneration^{53 54}. Furthermore, the regeneration and repopulation of hepatocytes in zones 1 and 3 are influenced by hepatocyte proliferation in zone 2. While the question regarding the origin of hepatocyte proliferation is ongoing, there is consensus that fully mature hepatocytes serve as the source of tissue replenishment, rather than a population of stem cells. The capacity of mature hepatocytes, which have already acquired functional specialization, to reacquire functions characteristic of cells in different zones suggests that hepatocyte identity is malleable and not inherently fixed during development. This also implies that various factors within the hepatocyte microenvironment may contribute to instructing tissue compartmentalization.

Interestingly, intermittent fasting (IF) induces rapid hepatocyte proliferation, albeit limited to PC hepatocytes, driven by systemic FGF15 and local Wnt signaling. This study suggests that hepatocyte proliferation during fasting and subsequent re-feeding serves to reestablish a consistent liver-to-body mass ratio. This challenges the idea that liver tissue primarily remains quiescent unless subjected to chemical or mechanical injury, and that IF can influence adult hepatocyte proliferation⁵⁵.

Beta-catenin signaling is pivotal in liver zonation, but frequent activating mutations in β-catenin observed in HCC, lead to uncontrolled cellular proliferation¹⁷. How do hepatocytes effectively balance this dual role of β-catenin signaling? The E3 ubiquitin ligases ZNRF3 and RNF43 cooperatively restrain β-catenin signaling, preventing unchecked proliferation and preserving liver zonation. Deletion of ZNRF3 and RNF43 shifts PP hepatocytes’ metabolism, leading to tumor development⁵². Although PC hepatocytes are susceptible to transformation⁵⁶, considering the ability of hepatocytes to be reprogrammed, the possibility remains that hepatocytes in other lobular regions could also give rise to tumors⁵⁷. Furthermore, distant tumors influence hepatic metabolism and disrupt zonated functions⁵⁸. In turn, fatty liver enhances the production and release of hepatocyte-derived extracellular vesicles that promotes the progression of colorectal cancer liver metastasis⁵⁹. These studies have uncovered new modes of communication between hepatocytes and remote tumors, prompting the need for further investigation.

7. Conclusions and outlook

Recent advancements highlight the importance of cellular compartmentalization, cooperation, and communication in optimizing liver function at various levels, echoing Malpighi’s prescient observations from centuries ago. Liver zonation is shaped by a collaborative effort of cells that communicate through local gradients, without physical boundaries. In response to external cues, hepatocytes can be reprogrammed and adopt new functions, via organelle remodeling and membrane contact sites. Cellular diversity and communication synergistically optimize liver function and not surprisingly, are altered during disease. Dynamic liver zonation is vital for adaptive homeostasis and tissue health, heralding an exciting new era in hepatic cell biology. New technology can now be used to delineate distinct gene, protein, and metabolite expression patterns^{60 12 61 62}, identify extensive cell-cell communications⁶³, and discern sub-organellar variations⁴⁰. These tools can be used to effectively fill in the gaps in our present understanding of liver function and regeneration in both physiological and diseased states. Given the imminent progress in the field, it is important to reiterate the importance of studying hepatic cell biology within a living system, especially with respect to the PP-PC axis and responses to physiological cues. Closing the knowledge disparities between mouse models and the human liver lobule is essential. Further insights into hepatocyte reprogramming mechanisms, cell-cell communication and strategies to restore zonal functionality are crucial for combating liver disease.