Deciphering the underlying mechanism of liver diseases through utilization of multicellular hepatic spheroid models

Abstract

Hepatocellular carcinoma (HCC) is a very common form of cancer worldwide and is often fatal. Although the histopathology of HCC is characterized by metabolic pathophysiology, fibrosis, and cirrhosis, the focus of treatment has been on eliminating HCC. Recently, three-dimensional (3D) multicellular hepatic spheroid (MCHS) models have provided a) new therapeutic strategies for progressive fibrotic liver diseases, such as antifibrotic and anti-inflammatory drugs, b) molecular targets, and c) treatments for metabolic dysregulation. MCHS models provide a potent anti-cancer tool because they can mimic a) tumor complexity and heterogeneity, b) the 3D context of tumor cells, and c) the gradients of physiological parameters that are characteristic of tumors in vivo. However, the information provided by an multicelluar tumor spheroid (MCTS) model must always be considered in the context of tumors in vivo. This mini-review summarizes what is known about tumor HCC heterogeneity and complexity and the advances provided by MCHS models for innovations in drug development to combat liver diseases.

INTRODUCTION

Alcohol, hepatitis B and C, as well as nonalcoholic liver disease, cause episodic and chronic inflammation of the liver (1). Inflammation not only evokes necrosis and genetic alterations in hepatocytes but also leads to fibrosis and cirrhosis, which trigger alterations of the microenvironment and may further contribute to the development of hepatocellular carcinoma (HCC) (Fig. 1) (2, 3). HCC has a Fast Track designation by the Food and Drug Administration (FDA), as there are no approved treatments for hepatic fibrosis. The only treatments approved by the FDA for HCC are angiogenesis inhibitors and immune checkpoint inhibitors (4). These medications cannot cure this complex, deadly disease; (5, 6) hence, scientists are using an efficient, physiological drug discovery platform to provide better translation from in vitro to in vivo experimental results.

Fig. 1.

Potential chemoprevention strategies for liver disease. Proposal of a potential chemopreventive strategy following the onset of liver disease.

For cancer drug discovery, the predominant platforms comprise two-dimensional (2D) cell assays; however, these models do not effectively predict the efficacy of drugs in vivo. Therefore, the clinical utility of drugs that proved effective in vitro is low. Also, assays of liver function, including cell polarization, albumin synthesis, bile secretion, and urea synthesis, are better when they are performed in three-dimensional (3D) cell cultures that more accurately reflect the normal context of liver cells (7). Similarly, 3D cells provide a better model system for cancer drug discovery.

The interactions between the tumor and the tumor microenvironment (TME) play a critical role in cell differentiation, tumorigenesis, metastasis, and the efficacy of drug therapies. Therefore it was important to develop a 3D TME platform to discover new therapeutics for liver cancer drug discovery. These platforms are complex and expensive to produce; however, the recently developed MCTS models provide a new platform for high-throughput screening (HTS) of drugs to treat a variety of diseases, including cancers and infections. The development of 3D MCHS models has had a major impact on drug discovery and the identification of novel targets for the treatment of HCC and fibrosis. MCHS models should greatly increase the number of potential new drugs to treat HCC and reduce the time to drug discovery.

In this review, we describe the influence of components of the TME on tumorigenesis, fibrogenesis, and chemoresistance in HCC, as well as advanced strategies that exploit 3D multicellular spheroid models to identify novel cancer targets and therapeutics.

INFLUENCE OF COMPONENTS OF THE MICROENVIRONMENT OF HCCS

The liver comprises hepatocytes, which are specialized parenchymal cells, and non-parenchymal cells. The latter include hepatic stellate cells (HSCs), Kupffer cells, and sinusoidal endothelial cells. Liver fibrosis results from inflammation that causes epithelial-to-mesenchymal transition (EMT), endothelial-to-mesenchymal transition (EndMT), and polarization of macrophages. These changes then lead to an increase in fibroblasts, which in turn accelerate both fibrosis and HCC.

Hepatic stellate cells (HSCs)

HSCs, found in the liver’s perisinusoidal space, are the primary cell type associated with liver fibrosis, HCC, and cirrhosis (8, 9). HSCs, which comprise about 10% of liver cells, are found in the subendothelial space in the diseased liver (10). Quiescent HSCs are typically very sensitive to extracellular fibrotic stimuli that arise from tissue injury, hepatitis, or inflammation. Quiescent HSCs, which have an abundance of vitamin A in lipid droplets, are important for normal immune function (11). However, activated HSCs located at the site of an injury produce a temporary scar of the extracellular matrix (ECM). Although the ECM protects the liver from further damage, this also initiates fibrosis. Thus, the activation of HSCs is the key process in hepatic fibrosis, leading to cirrhosis and HCC (10, 12).

After an injury to the liver, signals that are secreted by immune cells activate the quiescent HSCs to form a myofibroblast-like cell (9). However, how cancer cells and HSCs interact as tumors progress is unknown. Factors that activate HSCs include nuclear and G protein-coupled receptors (GPCRs), as well as fibrogenic growth factors and cytokines. In addition, hedgehog signaling and innate immune signaling, as well as changes in metabolism, activate HSCs (10, 13). Similarly, in response to virus-induced liver injury, HSCs are activated and transform from quiescent cells to myofibroblast-type cells (9, 13).

Increased expression of α-smooth muscle actin (α-SMA), which is characteristic of HSC activation in tumors, correlates with a worse prognosis for patients with HCC. When the activated HSCs interact with HCC cells, they overexpress VEGF-α and matrix metallopeptidase 9 (MMP9), which leads to a proangiogenic TME. We demonstrated that including activated HSCs in an MCHS model increases the expression of α-SMA and ECM production, which is an important driver of fibrosis (12). In addition, genetically or pharmacologically inhibiting 11βHSD1, which regulates the levels of glucocorticoids, results in the activation of HSCs. This occurs in MCHS and in a fibrotic environment as a result of the downregulation of Col1A1, p-SMAD3, Snail, and α-SMA (14).

Yet another key factor in promoting fibrosis is transforming growth factor beta (TGF-β), a cytokine that increases HSC activation. Activated TGF-β binds to TGF-receptor 2 and is translocated to the nucleus, where it regulates the transcription of pro-fibrotic genes and activates HSC via MAPK p38, JNK, and ERK pathways. TGF-β activates HSCs by targeting and binding to these cells, and it phosphorylates SMAD3, which stimulates type 1 and II collagen synthesis in the ECM (10).

Liver sinusoidal endothelial cells

The phenotypic change of an endothelial cells to a mesenchymal cell, which is called the EndMT, is important in tumor formation, metastasis, and fibrosis (15, 16). EndMT is induced by stresses such as oxidative stress and hypoxia and certain pathways, including Notch signaling and Wnt/b-catenin. These signaling pathways are controlled by transcription factors ZEB1, ZEB2, Snail, and Slug, which are important in the EndMT (17, 18). The changes seen in EndMT are similar to those in EMT, which is the major driver of the phenotypic cell changes that lead to a mesenchymal cell. However, there is recent evidence that the changes associated with EndMT are also important for the plasticity of the TME in cancer (15, 16).

Endothelial cells are transformed into fibroblast-like cells by TGF-β, which is also important during the EMT process in the TME, as it promotes tumorigenesis and the development of fibrotic disease. Several factors secreted by tumor cells, including TGF-β2 and interleukin-1b, induce EndMT in tumors and induce inflammation in endothelial cells (15, 18).

In lung cancer, secretomes from lung cancer spheroids induce EndMT in human umbilical vein endothelial cells (HUVEC). Inhibition of GSK-3β facilitates the reversion of EndMT in the TME. In a xenograft mouse model that includes an MCTS model containing lung cancer cells, inhibition of GSK-3β reduces the volume of the lung tumor and has a synergistic effect on lung cancer therapy (19). In lung cancer spheroids, expression of hypoxia-upregulated protein 1 (HYOU1) increases as a result of factors secreted via the crosstalk between endothelial cells and lung cancer cells. Repression of HYOU1 represses malignancy and stemness and increases apoptosis and chemosensitivity in a lung cancer MCTS model (20). Although GSK-3β and HYOU1 are important in many cancers as regulators of the reversion of EndMT, their role in HCC is still unknown (19, 20).

Immune cells

Macrophages are critical to HCC development and pathogenesis because inflammation is closely associated with HCC, a cancer that results from various factors, including alcohol, hepatitis virus, or nonalcoholic steatohepatitis. Tumor-associated macrophages (TAMs), a well-characterized component of the TME, are important for malignant growth because they secrete several cytokines and regulate the immune response to HCC cells. An increase in the number of TAMs is associated with cancer cell growth, invasion, metastasis, and angiogenesis. Similarly, high levels of TAMs correlate with a worse prognosis in patients with HCC (21).

The development of chronic inflammation because of the high level of infiltration of macrophages destroys the internal environment of the liver, furthers cirrhosis and fibrosis, and contributes to HCC development and progression. Macrophages can have two polarization phenotypes, M1 and M2 that are based on their response to microenvironmental stimuli. In the HCC, TAMs are mostly polarized towards the M2 phenotype (21, 22).

In HCC, M2 macrophages are essential for a) cancer cell migration through the TLR4/STAT3 signaling pathway (23), b) the secretion of CXCL8 (24) and c) TIM-3 increases (25). Also, IL-6 derived from TAMs promotes the invasion/metastasis of cancer cells. The increase of M2 macrophage-mediated EMT in HCC induces activation of the NTS/IL-8 pathway, which supports tumorigenesis in the inflammatory TME of HCC (26).

Cancer associated fibroblasts (CAFs)

Tumors contain malignant cancer cells; however, these cells are surrounded by non-malignant stromal cells. There is crosstalk between malignant cancer cells and cancer-associated fibroblasts (CAFs), which function similarly to myofibroblasts in the healing of wounds. However, CAFs promote the growth of tumors as well as tumor invasion and metastasis by a) degrading ECM proteins that leads to the remodeling of the ECM and b) secreting growth factors and cytokines that promote metastasis (27).

CAFs, which likely result from the EndMT, comprise a heterogeneous population of diverse cell origins and may have various effects on the growth of tumors and metastasis (15). Moreover, the TME that is associated with CAFs is induced by tumor EndMT (15, 28).

CAFs can remodel ECM to gain cancer-supporting properties, and the function includes matrix degradation, deposition, and stiffening. The enhanced EMT process contributes to increased malignancy and progression of HCC (28).

The signaling pathways involving CAFs are complex and contribute to the development and progression of cancer cells. Multiple fibroblastic cell populations that undergo mesenchymal transition may develop into CAFs when inflammation is present. However, the activation of quiescent HSCs, which is key to the initiation of the tumor and fibrosis, is the primary source of CAFs. In a pathological state of the liver, HCSs may become CAFs with the loss of p62, thereby promoting the formation of HCC (29).

CAFs have a function in the induction of HCC stemness. The liver cancer stem cells (LCSCs) that participate in the CAF-LCSC feedback loop help to transform CAFs and increase their ability to proliferate. In turn, CAFs induce SHH/Hh, IL-6/STAT3, Wnt/b-catenin, TGF-b/SDF-1/PI3K, and HGF/cMET pathways to maintain HCC stemness (30).

CHALLENGES IN USING MODEL SYSTEMS TO RECAPITULATE LIVER DISEASE

HCCs are largely resistant to chemotherapy; hence we need good model systems to find better treatments. Although monoculture cell lines for a variety of cancers have provided valuable new information on the biology of cancer cells, such systems cannot provide critical information on the crosstalk between cancer cells and the TME. Monoculture systems can be easily produced and have the advantage of reproducibility, but they lack heterotypic cell-cell interactions and cannot reproduce the complex TME (31). The complex milieu of a tumor comprises the active and ever-changing collective TME network comprising components of malignant and non-malignant cells, the ECM, and various signaling molecules (32). We need a physiologically relevant model that incorporates both the complexity and the dynamic changes characteristic of the tumor milieu.

The MCTS model recapitulates the complex interactions that characterize HCC and its microenvironment, effectively reproducing the complexity and heterogeneity of the tumor to provide an effective system to study cancer cells and identify targets and drug treatments (8, 33). For cancer research, MCTS models serve as a transition between in vivo tumor models and in vitro cultured cancer cells (33, 34).

Originally, to create an in vitro system that mimicked the complexity of the tumor in vitro, the MCTS models contained HCC cells and stromal cells, which comprise endothelial cells, HSCs, and fibroblasts. However, the robustness of the MCTS model varies with different stromal cell types, and in the HCC-MCTS, it depends on the crosstalk between HCC cells and HSCs; co-culturing of HCC cells with HSCs increases the resistance to cisplatin, doxorubicin, and sorafenib, which are used to treat HCC (8).The reciprocal crosstalk between parenchymal cells and non-parenchymal cells in the spheroid system more efficiently induces transformation of HSCs and endothelial cells than a similar monolayer culture system (15).

Thus, an MCTS model should mimic the in vivo interactions of the EndMT, EMT, and liver fibrosis (16, 35). Further, MCTS models comprising cancer cells and HUVEC or other endothelial cells can better reveal possible antiangiogenic activity, which increases metastasis compared to in vitro monolayer cultures. It was shown that 3D co-cultures of patient-derived xenograft organoids or HCC cells with endothelial cells increase the expression of CXCL16, IL-8, and MCP-1, indicating effective HCC cell-endothelial cells interactions in the 3D model (36). Interactions in the in vitro coculture of peripheral blood mononuclear cells (PBMCs) with HCC cells and non-HCC hepatocytes increase antigen presentation by HCC hepatocytes and activate PBMCs, leading to apoptosis of activated CD8+ T cells. Despite a somewhat effective immune response against HCC cells, the HCC hepatocytes escape the immune response (37). The efficacy of penetration of forskolin or retinoic acid in MCTS models combined with the ability of anti-cancer drugs to induce apoptosis proteins in MCTS models suggests that the 3D MCTS models are better than 2D coculture systems for studying cancer physiology and for finding new treatments (35).

IDENTIFICATION OF NOVEL TARGETS IN MCHS MODELS

Proteomic analysis of HCC 2D and 3D culture models revealed ASS1 (Argininosuccinate synthase 1) as a possible target; however, ASS1 expression is higher in the 3D culture model compared to the 2D culture model because the endoplasmic reticulum stress response is upregulated (Fig. 2) (38). Proteins in the secretome play important roles in cell communication, signaling, and migration. The importance of finding biomarkers and effective therapies for cancers drives studies of the secretomes of cancer cells to identify and characterize a) diagnostic markers, b) prognostic markers, and c) potential targets for drugs and other therapeutics (39). To identify secretome proteins in HCC cells, secretomes are collected from 2D and 3D model culture cells, proteins are separated by gel electrophoresis, and proteins in each spot are identified by mass spectroscopy. For HCC spheroids, the 3D culture model provides more secreted proteins, and they are of higher quality than the 2D culture model (40). Secreted proteins from HCC spheroids can be functionally classified as signal transduction proteins, biosynthesis and metabolic enzymes, and cytoskeletal actin filament stabilization proteins, among others. Bioinformatics tools, such as the STRING protein interaction network analysis, can identify the cell signaling pathways in which these secretome proteins function.

Fig. 2.

Identification of novel targets in MCHSs. Proteomic analysis of HCC 2D and 3D culture models revealed various target proteins were identified to verify the clinical significance of the targets.

New targets for HCC can be identified using different spheroid models that express the various malignant characteristics of HCC, such as angiogenesis (41), invasion/metastasis (42), and fibrosis progression (43). Quantitative proteomics of a 3D HCC culture model that was optimized for invasion characteristics identified proteins MMP2, MMP9, CXCL12, and CXCR4 early in HCC invasion. Thus, the dynamically changing pattern of proteins identified as invasion progresses reveals new targets to block the invasion and metastasis of HCC (44).

Chen et al. studied two HCC 3D spheroid models with different capabilities for metastasis and found significant differences in the patterns of expression of proteins involved in adhesion and invasion for the two models. This suggests that these differences are responsible for the metastatic and malignant characteristics of the HCC spheroids (42).

The 3D MCTS models are now often used for in vitro testing of possible therapeutic drugs because the 2D culture models often demonstrate greater sensitivity to the drugs (40). In an HCC spheroid model using stromal and HCC cells, LX-2 cells functioned similarly to tumor-CAF interactions, and the system showed resistance to the drug sorafenib (45). Similarly, another 3D culture model showed low sensitivity to sorafenib (40, 46).

Precision medicine requires reliable protein biomarkers that can be used for the early diagnosis of cancer and as targets for cancer therapy. Proteomics of the secretome of cancer cells, i.e., the proteins secreted or released by cancer cells, provides an efficient method for identifying cancer biomarkers and/or therapeutic targets (47). Jeon et al. reported that HCC patients with higher serum levels of sorbitol dehydrogenase (SORD) had shorter recurrence-free survival times; therefore, this protein could serve as a biomarker to identify HCC patients better suited for surgery. A high serum level of both alpha-fetoprotein (AFP) and SORD was also associated with a poor prognosis (48).

UTILIZATION OF MCHS MODELS FOR DRUG DISCOVERY

For HTS approaches, the MCHS model requires an efficient technique for producing 3D spheroids in multi-well plates. Although 3D spheroids are complicated and expensive to make, we have developed and validated a highly reproducible MCHS system with homogenously sized single spheroids in multi-well plates. This system is suitable for HTS approaches for drugs to treat cancer and fibrosis and for personalized medicine (Fig. 3).

Fig. 3.

Utilization of MCHSs for drug discovery. MCHS is utilized in drugs to treat cancer and fibrosis and for personalized medicine.

Anti-cancer

Despite the increased funding for research on various cancers and on drugs to treat these cancers, there are many incurable cancers. New 3D tumor models may provide a path forward for in vitro drug screening to identify clinically useful drugs (49).

Thanapirom et al. described a coculture platform that uses a decellularized human liver 3D ECM scaffold to test for drugs to treat cancer and fibrosis. The bioengineered decellularized 3D scaffolds contained HCC cells and HSCs taken from cirrhotic and healthy human livers. These cells were grown for up to 13 days in single culture or coculture and were shown to be an effective platform for screening anti-cancer drugs (50). Lafnoune et al. used a 3D cell culture model to test the efficacy of whole venoms and venom subfractions from the Moroccan scorpion (Buthus occitanus) and the Moroccan cobra (Naja haje) against HCC cells (51, 52). Liao et al. used an in vitro 3D spheroid culture system to identify CUDC-907 as a possible anti-HCC drug, which suggests that it may also be valuable for screening other drugs (53). Hou et al. took a different approach, using a 3D culture system comprising cells embedded in droplets of collagen to test the efficacy of new chemotherapy drugs (54). In addition, Wang et al. produced MCTSs of uniform size using agarose hydrogel microwells. They demonstrated that this platform is accurate, efficient, and reproducible for determining the efficacy of candidate drugs to treat cancer, and they used the platform to show that parthenolide may increase the efficacy of inhibitors that act on the FGFR4 receptor (34).

MCTS models comprising a mixture of HCC and stromal cells can be used to create MCHS models that mimic the microenvironment of liver tumors; thus, they can be used to determine the role of TME in the development of liver tumors. Further, the 3D TME of the MCHS model provides an efficient platform for HTS of drugs to treat liver cancer. For example, a phenomic screen using an MCHS model identified NA⁺/K⁺-ATPase inhibitor drugs as new targets for increasing the sensitivity of HCC cells (55).

Anti-fibrosis

As an alternative to target-based approaches, there is growing interest in developing phenotypic assays for screening drugs. For example, for liver fibrosis (56), 3D coculture models can be used for HTS to identify antifibrotic drugs (57).

The phenotype-based MCHS model, which surmounts the problems associated with 2D cultures, reflects the in vivo microenvironment of fibrosis. In our previous study, we constructed, characterized, and tested an MCHS model that mimics the in vivo microenvironment in liver fibrosis, and we used it to screen for drugs that could treat liver fibrosis effectively. In antifibrotic drug development, the MCHS model provides HTS using phenotypes to discover novel targets and drugs. After MCHS-based screening, the efficacy of EMT and EndMT inhibition is verified through 2D assays, and candidate drugs can be confirmed using an animal model of the disease (35).

Personalized medicine

Personalized medicine is a rapidly emerging clinical field that exploits recent bioinformatics technologies to improve the prediction, diagnosis, treatment, and prevention of disease (58).

For example, Fong et al. manipulated the conditions of an in vitro 3D macroporous cellulose sponge platform to promote the cultivation of HCC-PDX (patient-derived xenograft) organoids derived from 14 HCC-PDX lines (59).

An MCHS model derived from an HCC patient allows patient-specific optimization of therapy. HCC typically develops from tissues chronically damaged by inflammation, fibrosis, and liver cirrhosis; thus, patients are often resistant to standard chemotherapy. Because HCC is heterogeneous, we need an in vitro system to study this disease. To this end, two HCC cell lines secreting hepatitis B virus (HBV) DNA from infected patients were established, and these cell lines were used to develop a chemosensitivity assay for patient-derived MCHS. To monitor the results of the interaction of cancer cells and stromal cells in MCHS patient-derived HCC cells, human HSCs, endothelial cells, and fibroblasts were used for 3D cocultures to screen for effective therapeutics for HCC (31).

These 3D models extend beyond HCC. Zhang et al. developed an integrated real-time glioma system by producing cerebral 3D organoids ex vivo and xenograft tumors in vivo from patient glioma cells and tissues. They summarized the histological features, the chemotherapeutic drug response to drugs, and the clinical progression of the organoids in comparison to the parental tumors (60). Imamura et al. reported differences in the effects of chemotherapy by comparing 2D cultures and 3D cultures of breast cancer cell lines. They measured differences in drug sensitivity, expression of caspases and Ki-67, and oxygen status between the two types of cultured cells. The three 3D cell lines that form compact multicellular spheroids have increased resistance to doxorubicin and paclitaxel than the 2D cell lines (61). Kaur et al. developed a pancreatic adenocarcinoma cell line that was derived from a patient to study the complicated structure of the spheroid and the response of the model to therapeutic drugs in a 96- or 384-well format appropriate for robust performance in an HTS (62).

Infectious disease

Recently, there have been reports of 3D cell cultures for the development of a treatment for COVID-19 (63). Although mechanisms of infection have been studied in animal models as well as in 2D cell cultures, there are few organ-specific models of pathogenesis, especially for humans, which would require a mimic of a human organ. There are now engineered 3D models of human tissues that mimic many aspects of the physiology of human organs, including the human origin of the cells, the 3D structure, the response to physiological stimuli, and the tight cell-cell interactions (63, 64). Further, de Dios-Figueroa et al. reported that 3D cell culture models mimic major in vivo tissue properties such as virus-host interactions. These systems respond rapidly to new viruses, show reliable pathophysiological responses, and can be used to evaluate therapeutic drugs in a pandemic situation (65).

Salgueiro et al. studied 3D lung organoids as a model for lung pathogenesis and screening for therapeutic drugs. Human 3D lung organoids mimic the cell composition and physiological characteristics of the lung. For infections, 3D organoids provide a means to characterize pathogens and determine the response to infection. They demonstrated that lung organoids that were derived from cancer cells show greater susceptibility to infection by the influenza A virus with a reduced innate immune response compared to organoids from healthy tissue (66, 67).

CONCLUSION

HCC cells, along with the TME, rapidly acquire resistance to chemotherapy because the TME contributes heavily to both malignancy and chemoresistance. One approach to treating liver cancer is to target the components of the HCC microenvironment under physiologically relevant conditions. There have been various attempts to mimic the physiology of the in vivo environment, including 3D coculture systems. The WHO classification of digestive system tumors (5th edition) states that about 35% of HCCs can be described as histopathological variants with distinct molecular characteristics. Therefore, 3D coculture systems must be developed to reflect the unique characteristics of each histopathological variant of HCCs if these 3D models are to serve effectively in the management of HCC in patients.

Funding Statement

ACKNOWLEDGEMENTS This work was supported by the National Research foundation of Korea (NRF) grant (NRF-2022R1A2C1005371) and Korea National Institute of Health (KNIH) 2022-ER-1304-01 grant funded by the Korea government.