The adult zebrafish liver is structurally different from the mammalian liver. It contains three liver lobes, two dorsal lobes, and a single ventral lobe flattened along the intestine (Unterweger et al., 2023). Unlike the 2/3 PH in mammals, the resection of any two liver lobes led to intense bleeding, disrupted blood circulation, and subsequent mortality (Kan et al., 2009). Thus, the zebrafish PH model is characterized as a 1/3 partial hepatectomy in which the entire ventral lobe is surgically removed (Oderberg and Goessling, 2021).

After the PH surgery, the L/B ratio would be immediately reduced to approximately 65% of uninjured fish, indicating that the injury model is a 1/3 hepatectomy (Figure 2). The liver mass would be recovered to the average level 7 days post-surgery, as reflected by the L/B ratio. Over the next 2–3 days, an additional increase of L/B ratio (10% higher than normal fish) would be observed. After this, the L/B ratio decreased to 90% at 14 days post-surgery and then normalized at 28 days post-surgery (Kan et al., 2009). This pattern of liver regeneration in zebrafish is similar to mammalian liver regeneration, marked by an initial increase in liver mass, followed by a slow return to the original liver mass (Michalopoulos and DeFrances, 1997). Despite the L/B ratio presenting gender differences, males and females show similar regenerative dynamics after surgery (Kan et al., 2009).

Like the mammalian PH model (Michalopoulos and Bhushan, 2021), liver regeneration in the zebrafish PH model is achieved by the contribution of uninjured hepatocyte proliferation. Lineage tracing data reveal that BECs fail to contribute to hepatocyte regeneration after PH in zebrafish (Zhang et al., 2021), indicating the conserved regenerative process between zebrafish and mammals. However, zebrafish liver regeneration after PH could be divided into two ways: local regeneration and compensatory regeneration.

After 1/3 PH surgery in zebrafish liver, the regeneration of the missing ventral lobe could result in compensatory regeneration in the dorsal lobes other than the ventral lobe and ultimately lead to the recovery of liver mass within a week (Kan et al., 2009; Oderberg and Goessling, 2021). The compensatory growth of the liver after hepatectomy of the ventral lobe occurs in zebrafish, and this process, similar to those in rodents and humans, is closely associated with the activation and proliferation of hepatocytes (Kan et al., 2009). Transcriptomic profiling of the dorsal lobes following resection of the ventral lobe revealed significant changes related to compensatory regeneration (Feng et al., 2015). Meanwhile, another research group reported that the ventral lobe could mostly regenerate over 36 days post-surgery (Sadler et al., 2007), suggesting that the zebrafish owns the capacity for epimorphic liver regeneration, which differs in mammals (Goessling and Sadler, 2015). Considering the liver regeneration models vary with the extent of the injury, the technical variation in the PH protocol between research groups may lead to discrepancies in results.

In mammals, several signaling pathways have been shown to regulate PH-induced liver regeneration, including BMP, FGF, and Wnt signaling (Forbes and Newsome, 2016), which are also involved in regulating liver development (Zong and Stanger, 2012). Similar results were carried out using the zebrafish PH model. By using the zebrafish transgenic lines Tg(hsp:dnBPMR-GFP) and Tg(hsp:dnFGFR1-GFP), BMP and FGF were proved to be essential for hepatocyte proliferation after PH in both males and females (Kan et al., 2009). The Wnt/β-catenin pathway was shown to be activated after PH (Kan et al., 2009), implying that Wnt signaling also regulates liver regeneration in zebrafish. Indeed, Wnt activation by Wnt8a overexpression would promote PH-induced liver regeneration in zebrafish, while Wnt inhibition by dnTCF expression would reduce the regenerated liver mass (Goessling et al., 2008). Because PH-induced liver regeneration in zebrafish is achieved by hepatocyte proliferation, the modulation of cell cycle regulators could affect the regenerative process. Uhrf1 regulates the outgrowth of the developmental liver by inducing the genes involved in the cell cycle. Besides, uhrf1 heterozygous mutants exhibited defective regeneration after PH, indicating the essential roles of Uhrf1 in regulating the cell cycle upon acute liver injury (Sadler et al., 2007). Top2a has traditionally served as a marker for proliferation in both normal and cancer tissues, and Uhrf1 is recognized as a known positive regulator of Top2a activity (Wang et al., 1997; Hopfner et al., 2000; Zhu et al., 2002). Another study reported that top2a heterozygosity also caused the deficiency of liver regeneration in adult zebrafish (Dovey et al., 2009), suggesting that the promotion of Uhrf1-Top2a axis in cell proliferation is essential for liver regeneration in the PH model. The nucleolus complex Def-Capn3 also participates in PH-induced liver regeneration in zebrafish. Haploinsufficiency of def activates p53-dependent TGF-β signaling and causes scar formation after PH (Zhu et al., 2014), and Def-Capn3 complex the cell cycle reentry of hepatocytes by inhibiting Chk1 and Wee1 during liver regeneration (Chen et al., 2020). In addition to these factors, s-nitrosoglutathione reductase (GSNOR) plays opposing roles in PH-induced zebrafish liver regeneration, which was confirmed by the promoted phenotype upon genetic mutation or pharmacological inhibition of GSNOR (Cox et al., 2014).

The regulatory factors revealed in the zebrafish PH model deepen our understanding of hepatocyte proliferation upon hepatectomy (Figure 2). However, most of these findings have been previously reported in mammals, thus limiting the novelty of these studies. Additionally, due to the circulating breed system, the zebrafish PH model could hardly be used for drug treatment and screening. Hence, zebrafish is not a popular model to study PH-induced liver regeneration.

Based on the Mtz/NTR system, the transgenic lines Tg(fabp10a:Dendra2-NTR), Tg(fabp10a:mCherry-NTR), and Tg(fabp10a:CFP-NTR) were generated by two research groups (Choi et al., 2014; He et al., 2014; Choi et al., 2015). The hepatocytes of these three lines express fluorescent proteins that could be observed by fluorescent microscopes, and the hepatocytes can also be ablated by Mtz treatment (Figure 3). After 10 mM Mtz treatment for 24 h, nearly all hepatocytes would suffer from apoptosis, leading to extreme liver injury in zebrafish. In larval zebrafish, the injured livers could be functionally recovered at 48 h post Mtz treatment, making the larval zebrafish an ideal model to study liver regeneration (He et al., 2014). The Mtz/NTR system also works in adult zebrafish, and only 10 h of Mtz treatment could cause extreme hepatocyte ablation and functional regeneration can be achieved 5 days post-injury (He et al., 2014).

In addition to the above three commonly used transgenic lines that could induce extreme hepatocyte injury, another research group reported a new Mtz/NTR-based double transgenic model that could trigger moderate liver injury. This model contains a driver line Tg(fabp10a:GAL4-VP16,myl7:Cerulean) and a effector line: Tg(UAS:NTR-mcherry) (Jagtap et al., 2020). By crossing these two single lines with each other, hepatocyte-specific expressed GAL4 would activate the expression of UAS downstream genes (Halpern et al., 2008), leading to the hepatocyte-specific expression of NTR. In this model, even 2.5 mM Mtz treatment for 3 h would cause lethality of adult zebrafish. Thus, 1.5 mM Mtz treatment for 3 h was performed to induce hepatocyte injury. A large number of pyknotic nuclei and enucleated cells were detected at 3 h post-injury, suggesting that the hepatocytes were indeed suffering the injury. Unlike the extreme hepatocyte injury model, this model should be categorized as moderate to intermediate. The expression of hepatocyte marker fabp10a could always be detected at high levels during the regeneration process (Jagtap et al., 2020), indicating the moderate damage induced by Mtz/NTR system in this model.

Unlike hepatocyte proliferation-mediated liver regeneration, hepatocyte regeneration upon extreme liver injury is contributed by BEC transdifferentiation, which is validated by the Cre/Loxp-based lineage tracing system (Choi et al., 2014; He et al., 2014). The BEC-mediated liver regeneration consists of three steps: BEC dedifferentiation to bipotential progenitor cells (BPPCs), BPPC proliferation, and BPPC redifferentiation to hepatocytes and BECs. Firstly, extreme loss of hepatocytes leads to alterations of BEC morphologies and induction of the hepatoblast markers such as hhex and foxa3, indicating the dedifferentiation of BECs to BPPCs. Then, the BPPCs rapidly proliferate, which could be detected by EdU and PCNA staining. Lastly, BPPCs re-differentiate into nascent hepatocytes, which express the mature hepatocyte markers such as gc, bhmt, and tfa (He et al., 2019; Cai et al., 2021; Cai et al., 2023). In adult zebrafish, hepatocyte regeneration is also achieved by BEC transdifferentiation (Choi et al., 2014; He et al., 2014; Oderberg and Goessling, 2023), indicating the conserved regenerative process between larval and adult zebrafish.

In the moderate hepatocyte injury model Tg(fabp10a:GAL4-VP16,myl7:Cerulean; UAS:NTR-mcherry), hepatocytes but not BECs may play the major roles in liver regeneration. Because lineage tracing data were lacking in the model, all the conclusions made by the authors were based on the RNA-seq data (Jagtap et al., 2020). Even though some BEC markers showed induction at 12 h after injury, the changes were not significant, indicating that BECs would not rapidly proliferate upon hepatocyte injury. On the other hand, the continuous high levels of fabp10a also suggested that not all hepatocytes were injured, and the proliferation of uninjured hepatocytes could contribute to normal liver regeneration (Jagtap et al., 2020). However, the contribution of BEC transdifferentiation to liver regeneration could not be absolutely excluded, and the application of lineage tracing experiments is the only way to determine the exact sources of hepatocyte regeneration in this model.

Optogenetic approaches offer unprecedented precision in spatially and temporally regulating tissue manipulation, enabling minute control over biological processes (Portugues et al., 2013). To create a chemoptogenetic hepatocyte ablation tool for extended live imaging, the transgenic line Tg(fabp10a:dL5**-mCer3), abbreviated Tg(LiverZap), was generated. In this model, the exposure of 12 min NIR light would induce cell death of hepatocytes by producing ROS in larval zebrafish. Eight hours after the injury, apoptotic activity was observed in the liver region. Hepatocyte ablation was categorized into two distinct types: mild (30%) and severe (70%). Mildly ablated livers regenerated within 3 days post-injury, while severely ablated livers recovered within 7 days. Notably, the mechanisms of liver regeneration differed between the two conditions. Mildly ablated livers regenerated primarily through hepatocyte proliferation, whereas severely ablated livers were repaired via transdifferentiation of BECs. Additionally, using the LiverZap tool, the researchers demonstrated that targeted ablation of hepatocytes in a discrete region of interest is unexpectedly effective in triggering BEC-mediated regeneration, thereby challenging current perspectives on liver progenitor cell activation (So et al., 2020). Dynamic rearrangement of the biliary network and E-cadherin re-localization could both be observed in this model, indicating that cell adhesion modulation may be a pivotal step in BEC-mediated liver regeneration (Ambrosio et al., 2024). This model expands the current regeneration toolkit and enables detailed analysis of critical cellular dynamics, which are essential for understanding how the liver’s complex architecture is deconstructed during injury and reconstructed during repair. However, due to the variability in regeneration speed and dependency on injury severity, this model is limited in its ability to investigate the key factors involved in hepatocyte proliferation or BEC-mediated liver regeneration.

To identify small molecules that can facilitate liver progenitor cell (LPC)-to-hepatocyte differentiation, the Tg(fabp10a:pt-β-catenin) zebrafish transgenic line was constructed for LPC-mediated liver regeneration in which a mutated, stable form of Xenopus β-catenin is overexpressed in hepatocytes (So et al., 2021). The pt-β-catenin variant includes four point mutations (S33A, S37A, T41A, and S45A) at putative phosphorylation sites, and these mutations activate β-catenin by preventing its phosphorylation and subsequent degradation (Evason et al., 2015). Zebrafish hepatocyte-specific overexpression of pt-β-catenin would cause hepatocellular carcinoma (HCC) and recapitulate the pathologic features of human HCC (Evason et al., 2015). As early as 7 dpf, the hepatocytes of Tg(fabp10a:pt-β-catenin) larvae exhibited DNA damage, apoptosis, and senescence, which would be partially recovered at 30 dpf. The lineage tracing data showed that liver regeneration was achieved by both hepatocyte and BEC contributions. Thus, this model differs from the Mtz/NTR model. Using this model, the EGFR-ERK-Sox9 axis was found to play opposing roles during LPC-mediated liver regeneration, and the treatment of EGFR and ERK inhibitors would accelerate the differentiation of hepatocytes from LPCs (So et al., 2021). The inhibitory role of Sox9 in hepatocyte differentiation has also been observed in BEC-mediated liver regeneration, suggesting a conserved function across different models. Notably, EGFR and ERK signaling pathways exhibit opposing roles in BEC- and LPC-mediated liver regeneration (Cai et al., 2021; So et al., 2021; Oderberg and Goessling, 2023), highlighting the distinct regenerative mechanisms employed in different liver injury models. Additionally, PPARα activation is found to augment the differentiation of LPCs to hepatocytes by suppressing YAP signaling in this model (Kim et al., 2023).

A recent paper published in Development reports a new liver injury model in which liver cryoinjury is induced by adapting the CUBIC tissue-clearing approach in adult zebrafish. The ventral lobe was selected for cryoinjury due to its surgical accessibility. To ensure a reproducible and consistent injury, all procedures were performed at the level of the anterior fins and towards the midline. By 14 days post-cryoinjury, the injured area was nearly fully repaired. The cryoinjury would induce a localized necrotic and apoptotic lesion characterized by inflammation and infiltration of innate immune cells. Following the initial phase, the liver would suffer from fibrosis, which then be resolved by liver regeneration within 30 days. Cryoinjury would induce both localized and distal compensatory hyperplasia through trigger cell proliferation. The transcriptional landscape following cryoinjury has also been discovered (Sande-Melon et al., 2024). However, despite this, more information about this model needs further investigation, especially the functional validation of the regulatory factors.

After discovering BEC-mediated liver regeneration in zebrafish, this type of liver regeneration was also found in several mouse liver injury models. Three years after the zebrafish studies were published, by performing hepatocyte-specific knock-out of β1-integrin or overexpression of p21 combined with DDC/CDE/MCD-induced liver injury in mice, 15%–25% of the newly regenerated hepatocytes were found to be derived from BECs. However, the total experimental periods of this model reach about 2 months, and the contribution of BECs to hepatocytes is relatively low (Raven et al., 2017). Using the TAA/DDC-induced chronic liver injury model in mice, another group reported that chronic liver injury would also generate BEC-derived hepatocytes. Upon chronic injury for 6 months, about 10% of new hepatocytes were derived from BECs. These BEC-derived hepatocytes own a high proliferating capacity, and the ratio of these cells would reach 55% upon liver injury for 13 months (Deng et al., 2018). Besides, hepatocyte-specific deletion of β-catenin in mice could also trigger the BEC-to-hepatocyte transdifferentiation upon CDE-induced liver injury for 2 weeks. Statistically, about 20% of hepatocytes come from BEC transdifferentiation after liver recovery for 2 weeks, and this ratio reaches 70% after 6 months of recovery, confirming the proliferative capacity of BEC-derived hepatocytes (Russell et al., 2019b). Long-term CCl₄ treatment could also trigger BEC-derived hepatocyte regeneration in mice. However, like all the other models above, the contribution ratio of BECs is low, only up to 13% after 4 weeks of recovery upon CCl₄-induced injury for 16 weeks (Manco et al., 2019). Thus, these mouse models have similar disadvantages, including long experimental periods, operating difficulty, and low contribution ratio.

Compared to the mouse models, almost all newly regenerated hepatocytes are attributed to the BEC transdifferentiation in the zebrafish model (Choi et al., 2014; He et al., 2014). The Mtz/NTR system ensures extreme hepatocyte loss in zebrafish, thus providing a more specific model to study BEC-mediated liver regeneration. Additionally, the zebrafish larvae at 5 dpf could be used to induce liver injury, and the regeneration process can be completed as quickly as 48 h after Mtz treatment (He et al., 2014; Cai et al., 2023). Therefore, the experimental period of the zebrafish model is much shorter than that of the mouse models. This advantage provides the zebrafish BEC-mediated liver regeneration model as ideal for drug screening and regulatory factor exploration.

The process of liver regeneration via hepatocyte proliferation seems to be conserved, as the majority of hepatocytes in the resected lobe of zebrafish begin to proliferate within 2–3 days following partial hepatectomy (Kan et al., 2009). Research in zebrafish has identified multiple mechanisms that control liver regeneration, which is similar to mammals. However, the difference exists between zebrafish and mammals. In mammals, the hepatocytes have been proven to be heterogeneous by several genetic lineage tracing and single-cell sequencing studies (Aizarani et al., 2019). By using lineage tracing from the telomerase reverse transcriptase (Tert) locus in mice, rare hepatocytes with high telomerase expression are found to be distributed throughout the liver lobule. Upon injury, the repopulating activity of these Tert^high hepatocytes accelerates. When Tert^high hepatocytes are genetically ablated in combination with chemical-induced liver injury, there is a marked increase in stellate cell activation and fibrosis (Lin et al., 2018), suggesting the contribution of these cells in liver regeneration. Additionally, Axin2⁺ pericentral hepatocytes may contribute to liver homeostasis and repair (Wang et al., 2015) despite the controversial conclusion (Sun et al., 2020). These studies adequately prove the heterogeneity of hepatocytes in mammalian livers. However, no evidence has been reported supporting the hepatocyte heterogeneity in zebrafish. Besides, a systematic analysis of zebrafish liver is lacking to compare the difference between zebrafish and mammals.

Compared to hepatocytes, BECs exhibit similar plasticity in zebrafish, mice, and humans. Upon extreme or chronic hepatocyte injury, BECs similarly have the capacity to transdifferentiate into functional hepatocytes (He et al., 2014; Raven et al., 2017; Gribben et al., 2024). The regulatory factors involved in BEC-mediated liver regeneration seem to be conserved cross species. Wnt signaling controls the redifferentiation of BPPCs into hepatocytes in both mice and zebrafish (Huang et al., 2014; Pu et al., 2023), while Notch signaling also governs the redifferentiation of BPPCs into BECs in mice and zebrafish (Ko et al., 2019; Cai et al., 2021; Pu et al., 2023). Similar roles of FXR, Hdac1, VEGF, Mdka, and BET have been proved in mouse and zebrafish BEC-mediated liver regeneration models (Ko et al., 2016; Ko et al., 2019; Cai et al., 2021; Cai et al., 2023; Rizvi et al., 2023; Zhang et al., 2024). Besides, the PI3K-AKT-mTORC1 pathway regulates the BEC-to-hepatocyte plasticity in both zebrafish and humans (He et al., 2019; Cai et al., 2023; Gribben et al., 2024).

Structural and cellular differences in the liver between mammals and zebrafish may lead to distinct regenerative responses. In zebrafish, the bile duct structure is defined by a single layer of cuboidal cells, which is specific to hilar bile ducts and not present in peripheral bile ducts (Cai et al., 2021). The lumen formation of peripheral bile ducts typically occurs through the fusion of cytoplasmic vesicles between adjacent biliary cells (Lorent et al., 2010), a process distinct from that in mammals. This unique structure of the zebrafish liver may account for the rapid response of BECs to hepatocyte injury, potentially explaining the high frequency of BEC-mediated liver regeneration observed in zebrafish injury models (He et al., 2014; Ambrosio et al., 2024). Furthermore, zebrafish lack the liver-resident macrophage Kupffer cell, which is found in mammals (Goessling and Sadler, 2015). In mammals, Kupffer cells secrete pro-inflammatory cytokines upon liver injury, which are crucial for hepatocyte proliferation and liver regeneration (Wang et al., 2024). The absence of Kupffer cells in zebrafish may explain the lack of inflammatory factors involved in BEC-mediated liver regeneration and could influence the preferred pathway for liver regeneration.

Liver transplantation is the only way to treat end-stage liver diseases, which have high mortality (Fisher, 2017). However, the shortage of organ donors and graft rejection limit the application of liver transplantation to many patients (Fisher, 2017). In patients with end-stage liver diseases, hepatocyte proliferation is compromised due to liver steatosis, inflammation, fibrosis, and cirrhosis (Stanger, 2015). Thus, the facilitation of hepatocyte proliferation-mediated liver regeneration seems to be infeasible for liver repair in these cases. Ductular reaction, reflected by BEC proliferation and progenitor activation, is the hallmark of almost all chronic and acute liver diseases (Sato et al., 2019). After the findings of BEC-mediated liver regeneration, promoting the BEC-to-hepatocyte transdifferentiation was postulated to be the alternative way to alleviate severe hepatocyte injury in end-stage liver diseases.

Using the zebrafish extreme liver injury model, VEGF signaling was found to be essential for BEC-to-hepatocyte transdifferentiation (Cai et al., 2023; Rizvi et al., 2023). Based on this phenomenon, another group explored the clinical benefit of VEGF activation in promoting liver repair and restoring liver function in several mammal liver injury models. Using the mRNA-lipid nanoparticles (LNP) delivery system, the controllable transient VEGFA expression in the liver could be archived (Maestro et al., 2021). VEGF mRNA-LNP robustly induced the BEC-to-hepatocyte transdifferentiation and promoted the recovery of liver function in both chronic and acute mouse liver injury models (Rizvi et al., 2023), suggesting the potential clinical benefits of VEGFA mRNA-LNP to alleviate liver diseases. In the livers of patients with end-stage liver diseases, intermediate hepatocyte-like cells exist and express both hepatocyte and BEC markers. Moreover, the presence of these cells is associated with the expression of VEGF receptor KDR (Rizvi et al., 2023), implying that VEGF activation may also stimulate BEC-mediated liver regeneration for human liver disease intervention. Additionally, even though there is a lack of data from mammals, the supplement of either 17 β-estradiol (Chaturantabut et al., 2019), FGF21 protein (Qiang et al., 2021), or PPARα agonist (Kim et al., 2023) could promote the transdifferentiation of BECs to hepatocytes in zebrafish. These findings suggest that using zebrafish as the model of BEC-mediated liver regeneration could indeed provide a potential translational impact on human health.

Although significant progress has been made in studying liver regeneration using zebrafish models, a comprehensive understanding of hepatic biology and disease pathology requires the integration of diverse research tools. The development of liver organoids, a self-organizing and self-renewing three-dimensional cell culture model, has significantly advanced liver research. Liver organoids offer a physiologically relevant platform for drug screening and development, personalized medicine, disease modeling, and the study of liver regeneration (Chawla and Das, 2023). Recently, liver organoids have been used to alleviate liver damage and promote liver repair in mice (Aloia et al., 2019; Yuan et al., 2024), indicating the huge therapeutic potential for liver diseases. However, the emergence of organoid technology is accompanied by several limitations that must be addressed, including the high cost of these models, limited availability of source tissues, and the need for multilineage liver organoids to accurately replicate the cellular heterogeneity of the liver (Chawla and Das, 2023). In contrast, zebrafish offer several advantages, including low cost, high reproductive capacity, and the ability to replicate an in vivo microenvironment. Therefore, considering the distinct benefits of zebrafish models and organoids, integrating these tools is recommended for studying liver regeneration, particularly in pre-clinical research.