Engineering liver microtissues for disease modeling and regenerative medicine

Abstract

The burden of liver diseases is increasing worldwide, accounting for two million deaths annually. In the past decade, tremendous progress has been made in the basic and translational research of liver tissue engineering. Liver microtissues are small, three-dimensional hepatocyte cultures that recapitulate liver physiology and have been used in biomedical research and regenerative medicine. This review summarizes recent advances, challenges, and future directions in liver microtissue research. Cellular engineering approaches are used to sustain primary hepatocytes or produce hepatocytes derived from pluripotent stem cells and other adult tissues. Three-dimensional microtissues are generated by scaffold-free assembly or scaffold-assisted methods such as macroencapsulation, droplet microfluidics, and bioprinting. Optimization of the hepatic microenvironment entails incorporating the appropriate cell composition for enhanced cell-cell interactions and niche-specific signals, and creating scaffolds with desired chemical, mechanical and physical properties. Perfusion-based culture systems such as bioreactors and microfluidic systems are used to achieve efficient exchange of nutrients and soluble factors. Taken together, systematic optimization of liver microtissues is a multidisciplinary effort focused on creating liver cultures and on-chip models with greater structural complexity and physiological relevance for use in liver disease research, therapeutic development, and regenerative medicine.

Graphical Abstract

1. Introduction

The liver is the largest internal organ and the primary site of metabolism in the human body, performing a myriad of essential functions including: synthesis and metabolism of carbohydrates, proteins, and lipids; detoxification of xenobiotics; fluid and electrolyte balance; bile secretion; and inflammatory response regulation.^[1] These processes are impacted by each individual’s genetic makeup, medical history, lifestyle, and environment, constantly adapting to new demands imparted on the liver. To withstand these demands, the liver is protected by a phenomenal regenerative capability, which is tightly regulated by coordinated signaling among different resident cell types.^[2]

A variety of liver diseases such as hepatitis (viral and nonviral), metabolic disorders, cancer, alcoholic and non-alcoholic fatty liver diseases, can result in hepatic dysfunctions, homeostasis disruptions, and organ failure. Furthermore, drug-induced liver injury (DILI), the leading cause of post-marketing drug withdrawals, can also lead to acute liver failure and death.^[3] In fact, the global burden of liver disease is increasing, accounting for approximately two million deaths annually and resulting in large economic burden and low quality of life.^[4] For patients with severely impaired liver functions, transplantation is the only effective treatment, which unfortunately, is limited by the shortage of available, healthy donor organs. Cell therapy approaches are being developed as an alternative to whole organ transplant and hold tremendous potential to treat many metabolic, acute, and chronic liver diseases.

Cellular models for liver disease and regeneration mainly focus on hepatocytes, the major parenchymal cells that are the workhorse of metabolism and make up 80% of the liver mass.^[1b] In the past decade, tremendous efforts have been applied to deriving high-quality hepatocytes from cell sources other than primary liver explants, such as pluripotent stem cells ^[5] and other adult tissues.^[6] More complex 3D models that better mimic liver physiology have been developed by incorporating other resident liver cell types together with biomaterials.^[7] These 3D aggregates, or “microtissues”, have shown improved cellular organization and hepatic functions versus 2D monocultures,^[8] and though many major challenges remain, they have already provided valuable insights into hepatic diseases and regeneration.

This review describes bottom-up approaches to engineering 3D liver microtissues, and strategies to optimize the hepatic microenvironment and culture conditions to enhance the 3D architecture and functions of liver microtissues. Following this special issue’s theme on disease diagnosis and tissue regeneration, we will also explore the applications of liver microtissues in disease modeling and regenerative medicine.

2. Evaluating Cell Sources for Liver Tissue Engineering

Selecting an appropriate cell source for a liver tissue engineering application is critical; considerations include cost, availability, reproducibility, scalability, and physiological relevance (Figure 1A). While each researcher must evaluate a model’s suitability for a specific research question, important functions of liver culture systems include their ability to synthesize and secrete hepatocyte-specific proteins (such as albumin, clotting factors, and α−1-antitrypsin); their ability to metabolize nitrogenous compounds to urea; and their ability to induce cytochrome P450 (CYP) enzyme expression in response to xenobiotics. Additional hepatocytic properties or functions that may be relevant include membrane polarity (distinct membranes expressing apical- and basolateral- specific markers ^[9]), bile acid synthesis, glycogen storage, and lipoprotein trafficking (Figure 1B).

Figure 1.

Assessment of hepatocyte sources for liver tissue engineering. A) Considerations for choosing a hepatocyte source for disease modeling or regenerative medicine. B) Schematic of hepatocyte polarity and summary of their essential functions (1–8); when optimizing liver microtissues, some of these functional outputs are measured and compared across different culture platforms or treatment groups. Parts of this figure created with BioRender.com.

2.1. Primary Human Hepatocytes

Primary human hepatocytes (PHH) are obtained directly from whole or partial liver tissue by using enzymatic digestion of the extracellular matrix (ECM) to dissociate parenchymal cells.^[10] These freshly isolated cells remain the closest representation of in vivo liver physiology and are widely accepted as the gold standard for in vitro investigation of xenobiotic biotransformation, enzyme induction and inhibition, hepatotoxicity, viral hepatitis, and liver regeneration.^[11]

Primary hepatocytes are terminally differentiated and therefore do not generally proliferate. Their scarcity and cost are major limitations for their use in whole tissue engineering and cell therapy. PHHs vary among donors due to genotypic and environmental factors and due to disparate sourcing within hepatic lobules, and therefore require thorough characterization for each new donor.^[12] PHH donor variability can be exploited for powerful disease modeling and mechanistic studies as many bio-banks now compile extensive genomic data on their inventory, allowing purchase of hepatocytes with specific mutations of interest.^[13] Since PHHs are scarce and costly, primary hepatocytes can also be obtained from non-human species as a more affordable alternative. Non-human hepatocytes are most often obtained from rodents, in part due to the ease of breeding from a common genetic background, which allows for improved experimental consistency; larger animals are useful when a greater number of cells is required. However, the ability of these animal liver cells to recapitulate human physiology is a concern for liver disease modeling, drug efficacy and hepatotoxicity prediction,^[14] and these cells also pose a risk for immunogenicity and xenozoonosis if used in transplantation.^[15]

2.2. Immortalized Hepatic Cell Lines

Hepatocytes derived from immortalized human hepatoma lines such as HepG2, Huh7, HepaRG, and C3A are well-characterized, and display polarized expression of membrane transporters and bile canalicular structures and other hepatocyte-specific properties in 3D cultures.^[16] The ability of immortalized hepatic cells to proliferate rapidly makes them more cost-effective than primary hepatocytes and an appealing cell source for large-scale tissue- engineering endeavors. Their derivation from a common genetic origin significantly enhances their reproducibility, relative to PHHs. However, these cell lines are often limited by the instability of their chromosomal profile and polyploidy,^[17] and from problems in 3D culture such as development of necrotic cores and loss of structural integrity.^[16c] HepG2 gene expression and secretome profiles more closely mirror those of fetal than adult hepatocytes, limiting their use.^[18] HepaRG cells exhibit high xenometabolic activity ^[19] and have been proposed as a surrogate for PHHs in hepatic disease modeling^[20].

2.3. iPSC-derived Hepatocyte-like Cells

Embryonic stem cells (ESCs) are pluripotent stem cells derived from the undifferentiated inner cell mass of a blastocyst or embryo. ESCs possess the ability to proliferate indefinitely and differentiate into any somatic cell upon appropriate stimulation,^[21] including hepatocyte-like cells (HLCs). ESC-derived HLCs express hepatocyte-related genes and mimic liver functions,^[22] and are able to colonize mouse liver tissue after transplantation and promote endogenous regeneration.^[23] However, due to practical and ethical concerns that preclude the clinical application of ESCs, induced pluripotent stem cells (iPSCs), have become a viable alternative to ESCs for creating HLCs.^[24] iPSCs are generated through the transcriptional reprogramming of adult somatic cells; the ease of access to donor cells from blood, skin, hair, or urine, makes iPSC-derived models a valuable tool for the advancement of precision medicine.

Protocols to generate stem cell-derived hepatocytes that resemble PHHs are constantly being improved, and many groups have produced HLCs that display liver-specific gene and protein expression, protein secretion, CYP enzyme metabolism and induction, and glycogen storage.^[25] Thus far, no protocol has yielded mature liver cells that are functionally equivalent to primary adult hepatocytes; stem cell-derived hepatocytes typically exhibit a phenotype closer to that of fetal hepatocytes.^[26] Various 3D culture approaches (reviewed below and elsewhere^[27]) have been employed to enable further maturation and improved functionality of HLCs over time. While these cells offer great potential for disease modeling and personalized cell therapy, several obstacles to the use of HLCs in clinical applications need to be addressed. Most importantly, HLCs as a hepatocyte source for regenerative or cell therapy pose a significant risk for teratoma formation due to iPSC contamination resulting from suboptimal differentiation efficiency (a weakness of current protocols).^[28] The long-term efficacy of these cells and the large-scale manufacturing with high reproducibility remain to be confirmed.^{[6c, 28a, 29]}

2.4. Induced Hepatocyte-like Cells from Other Adult Tissues

Various strategies have been devised to directly reprogram, or transdifferentiate, human somatic cells to create induced hepatocyte-like cells (iHeps). This approach maintains the translational benefits of HLCs while bypassing the pluripotent stage and thus mitigating the risk of tumorigenicity following implantation.^[30] Huang et al. were among the first groups to demonstrate that human adult fibroblasts could be converted into iHeps that display mature hepatocyte functions such as CYP enzyme activity and biliary drug clearance in vitro, and the ability to restore liver function and prolong mouse survival when transplanted into mice with acute liver failure in vivo.^[6a] Song et al. developed a mouse model in which liver-resident myofibroblasts were directly reprogrammed to iHeps capable of storing glycogen, secreting albumin, and expressing drug-metabolizing enzymes.^[31]

Methods for iHep production rely on overexpression of defined transcription factors, and investigators must weigh the risk of unintended insertional mutagenesis from lentiviral delivery of transgenes against the inefficiency of non-integrating viruses or other strategies (reviewed by Grath et al.^[6d]). It is important to note that despite achieving a reprogramming efficiency of only 4%, Song et al. still observed significant amelioration of chemical-induced liver fibrosis in their model, demonstrating the exciting potential of transdifferentiation techniques in cell therapy and treating chronic liver diseases.

3. Generating Liver Microtissues

In conventional 2D monolayer culture, PHHs survive for only about a week, during which they undergo morphologic alterations and loss of key hepatic functions.^[32] This process, termed dedifferentiation, has been attributed largely to disruption of the intricate cell-cell and cell-ECM interactions that are present in intact tissue.^[18a] To prevent dedifferentiation, a variety of 3D tissue engineering approaches are used to mimic the in vivo liver environment for functional maintenance of PHHs or maturation of HLCs and iHeps, improving their utility for studying liver diseases, regeneration, and hepatotoxicity.^{[8a, 8b, 33]}

Creating liver microtissues is a multidisciplinary endeavor, requiring expertise from engineering and materials science in addition to molecular and cellular biology. Hepatocytes possess incredible self-organizing ability in 3D culture, and the use of biomaterials can improve the structural and functional complexity of the microtissues. In this section, we will first discuss scaffold-free microtissue generation with an emphasis on the recent development of organoids, followed by research on scaffold-assisted approaches, including macroencapsulation, microfluidics, and bioprinting. Some of the technologies presented here offer controllable 3D tissue sizes and are used to produce spheroids on the millimeter scale, thus the word “microtissue” will be used to describe aggregates on both micrometer and millimeter scales.

3.1. Scaffold-free Liver Microtissues

Scaffold-free spheroids are among the simplest forms of engineered liver microtissues. Cell suspensions self-aggregate into spheroids due to a combination of gravitational, hydrodynamic, and electrostatic forces (Figure 2A).^[34] Arrayed platforms such as hanging drops ^[35] and micro-wells ^[36] have been used widely to improve the throughput and uniformity of 3D cell aggregates. Liver spheroids consisted of primary hepatocytes, hepatic cell lines, HLCs, or iHeps have all exhibited improved functions such as maturity and drug metabolism when compared to their 2D counterparts. For example, HepaRG spheroids formed in 96-well ultra-low attachment plates demonstrated dose-dependent DNA damage in response to genotoxic compounds and higher sensitivity compared with 2D culture.^[34b]

Figure 2.

Generation of liver microtissues. Scaffold-free assembly by: A) aggregation of mature hepatocytes, or B) self-organization of hepatic organoids. C) Single-cell analysis identified key cell types in liver organoids. Macroencapsulation by: D) prefabricating the scaffold before cell seeding, or E) encapsulating cells during gelation. Droplet-based microfluidics allows cell aggregation and manipulation of microtissues within: F) W/O/W droplets, or G) W/W/O droplets. H) Bioprinting can be used to construct hepatic structures such as lobules with high precision and specific organization. (A) Reproduced with permission.^[34] Copyright 2019, Springer Nature. (B) Reproduced with permission.^[42] Copyright 2013, Springer Nature. (C) Reproduced with permission.^[43] Copyright 2019, Elsevier. (D) Reproduced with permission.^[138] Copyright 2019, Elsevier. Reproduced with permission.^[47] Copyright 2011. Elsevier. (E) Reproduced with permission.^[49] Copyright 2010, Mary Ann Liebert. (F) Reproduced with permission.^[8] Copyright 2016, Wiley–VCH. (G) Reproduced with permission.^[56] Copyright 2016, The Royal Society of Chemistry. (H) Reproduced with permission.^[7] Copyright 2016, PNAS. Parts of this figure created with BioRender.com.

Organoids are multi-cellular structures that recapitulate some levels of tissue architecture and functional complexity, and they are derived from adult tissues or pluripotent stem cells.^[37] Self-assembled liver organoids have become another popular approach to generate microtissues with complex structural features. Current debate is inconclusive over the existence and identification of adult hepatic stem cells, but evidences suggest that hepatocytes and cholangiocytes (bile duct epithelium) have a high degree of cell fate plasticity in response to injury.^[38] Some groups have identified and cultured hepatic progenitors that are suitable for generating organoids.^[39] Clevers’ group isolated Lgr5+ cells which appeared near bile ducts of injured murine liver. They were mixed with Matrigel for clonal expansion and subsequently differentiated to hepatocytes.^[39a] This group further extracted EpCAM+ ductal cells from human liver, expanded them as progenitors and generated human liver organoids, which retained their long-term genetic stability.^[40]

iPSC-derived organoids were originally formed by aggregating iPSCs in suspension to make embryonic bodies (EBs), which were then exposed to morphogens to induce hepatic differentiation.^[41] However, this method was less efficient than directed differentiation to HLCs in 2D culture. Alternatively, Takebe et al. generated hepatocyte endoderm from human iPSCs, then reseeded the hepatocytes with mesenchymal and endothelial cells on a pre-solidified Matrigel surface and allowed the cells to self-organize into a 3D aggregate or “liver bud” (Figure 2B).^{[7b, 42]} These liver buds with nascent vasculature could be further matured after implantation into mice. A recent study by Takebe’s group embedded iPSC-derived foregut spheroids in Matrigel before induction to hepatic organoids, which encompassed key cell types found in the native liver, including hepatocyte-, biliary-, stellate-like cells (Figure 2C).^[43] The remarkable self-organizing ability of organoids makes them a great system to study organogenesis and morphogenesis in vitro. In a separate experiment, the fusion of iPSC-derived anterior and posterior gut spheroids allowed the coordinated development of hepato-biliary-pancreatic tissues at the foregut-midgut boundary.^[44]

3.2. Macroencapsulation

Macroencapsulation uses bulk materials as scaffolds to support liver microtissue formation and functions. Two popular macroencapsulation strategies are (1) direct seeding of cells into prefabricated scaffolds followed by spontaneous aggregation, and (2) combining cells with biomaterials and inducing gelation to enclose them (Figure 2D,E). In the first strategy, scaffold fabrication does not include the cells and may involve harsh solvents and reactants.^[45] However, the final product must be biocompatible and possess the appropriate porosity and topography to support cell aggregation. Scaffold architectural features such as fiber diameter, pore size, porosity, interconnectivity, and surface properties, have a direct impact on cell viability, proliferation, differentiation, and ECM response. Larger pore size and higher porosity allow for efficient gas and nutrient exchange and the formation of larger aggregates, but result in lower mechanical strength for the overall scaffold.^[46] One example of a prefabricated porous scaffold is a soft hydroxypropyl cellulose sponge conjugated with galactose, which interacts weakly with asialoglycoprotein receptors (ASGPR) on the hepatocyte membrane, promoting the formation of spheroids (Figure 2D, right).^[47] The macroporous, hydrophilic network had an average pore size of 110–130 μm and provided physical constraint for rat hepatocytes, which formed spheroids with an average diameter of 60.7 μm. In a follow-up study, the authors modified the cellulosic sponge with a disulfide-cleavable linkage during polymer synthesis.^[8a] HepaRG maturation was expedited in 3D, and mature spheroids were retrieved after adding reducing agents to the scaffold.

Encapsulation of hepatocytes during scaffold formation requires non-toxic polymers and mild gelation conditions. Hydrogels are attractive due to their biocompatibility, high water content, and tunable biodegradability; however, their gelation, degradation rate, and mechanical properties require more optimization.^[45] Prior to gelation, cells are mixed with a viscous hydrogel precursor which is crosslinked in response to changes in temperature, pH, light, or chemicals.^[48] The resulting 3D hydrogel network is porous and soft but provides sufficient physical support for the encapsulated cells. Zhao et al. mixed primary rat hepatocytes with a collagen type I solution, which was then adjusted to neutral pH to form a collagen gel (Figure 2E, right).^[49] The hydrogel was dissociated into cylindrical hepatic units that were 2–4 mm in diameter and 0.5–1 mm in height, and were subsequently implanted into the subcutaneous space in rats. The collagen hydrogel showed enhanced engraftment of hepatocyte tissues in vivo, with blood vessels connected to the host system within 3 days.

3.3. Droplet-based Microfluidics

Droplet-based microfluidic technology utilizes small amounts of liquid and reagents at the same scale as cells, making it attractive to the biological sciences. Capillary and chip-based systems have been developed to produce, process, and assay droplet populations, and these systems have great potential for use in optimizing the microenvironment for hepatocyte culture.^[50] Droplet-based microfluidic systems provide control of microtissue manufacturing parameters such as aggregate size, morphology, and composition,^[51] and allow high-throughput manipulation and production of hundreds of homogeneous, monodisperse droplets per minute.^[52] Cell encapsulation and subsequent aggregation within emulsion droplets require fine tuning of the system, including designing microfluidic channels and optimizing flow rates to maintain laminar flow, selecting reagents with low viscous shear stress and interfacial tension to ensure successful emulsions, and incorporating biomaterials for 3D encapsulation and maintenance.

Droplet-based microfluidic systems have been used to promote hepatocyte aggregation within hydrogels in large numbers of isolated droplets. In the inner phase, cells are mixed with hydrogel precursors that polymerize upon changes in temperature, UV exposure, or chemical composition.^[53] This inner aqueous phase is surrounded by an oil phase to create water-in-oil (W/O) single emulsion droplets. To further stabilize the emulsions, an outer aqueous phase can be added to encapsulate the oil phase to create water-in-oil-in-water (W/O/W) double emulsion droplets (Figure 2F).^[54] A few groups have applied the W/O/W principle to design microfluidic chips for generating liver microtissues.^{[8b, 55]} Chan et al. developed a double emulsion platform to encapsulate primary rat hepatocytes and endothelial progenitor cells within a hydrogel of collagen and alginate; the collagen gel provided functional support to the cells while the alginate gel provided physical support to keep the spheroids in suspension and prevent spheroid fusion.^[8b] The cells coalesced to form a single spheroid of 80 μm in diameter within each droplet in 4 hours.

In some applications, the cell/scaffold phase is divided into two distinct aqueous phases, resulting in water-in-water-in-oil (W/W/O) emulsions and core-shell structures (Figure 2G).^[56] In one study, the authors suspended HepG2 cells in growth medium (inner phase), fibroblasts in an aqueous alginate solution containing Ca-EDTA complex (middle phase), and co-flowed both through the microfluidic device without mixing.^[56b] An outer phase of fluorinated carbon oil was added which encapsulated volumes containing the inner and middle phase to create droplets. A second oil layer containing acetic acid was then introduced and triggered the release of Ca²⁺ from the Ca-EDTA complex, inducing gelation in the middle alginate phase. The final droplet consisted of a core-hydrogel shell structure in which hepatocyte spheroids grew to 169 μm in the center, surrounded by fibroblasts in the periphery.

3.4. Bioprinting

3D bioprinting can be used to produce 3D hepatic structures with high precision and specific organization.^[57] There are four bioprinting technologies commonly used for deposition and patterning of biological materials: inkjet printing, microextrusion, stereolithography, and laser-assisted printing.^[58] Each of these technologies has advantages and disadvantages. Microextrusion can accommodate high cell densities and even spheroids, whereas inkjet printing can accommodate only up to 10⁶ cells/mL.^[59] However, inkjet printing can operate at higher speed and lower cost than microextrusion, and requires shorter preparation time.^[60] Stereolithography is fast and nozzle-free, but the UV light used for photocuring can cause cellular damage.^[58] Laser-assisted printing can achieve the highest resolution but is more costly.^{[58, 60]}

Many natural, synthetic, and hybrid polymers are used as bio-ink for 3D bioprinting, with varying degrees of success. Collagen type I is commonly used as a hepatocyte scaffold, but it is not ideal for bioprinting due to its low viscosity and slow polymerization.^[61] A improved hybrid of methacrylate collagen type I and thiolated hyaluronic acid was developed as a bio-ink for microextrusion printing and UV polymerization.^[61] Using this hybrid, >80% PHH viability was observed and the hydrogel supported the cells for over two weeks. Photocrosslinkable, decellularized liver ECM has also been developed as a bio-ink to provide niche-specific signals.^[62] A flexible bio-ink formulation coupled with modular printing technologies can be used to print spheroids for creating larger microtissues,^[63] to construct organ-on-a-chip platforms,^{[63a, 64]} and to allow cell remodeling of the microenvironment.^[65]

Recapitulating the lobule, the functional unit of the liver, is of great interest in liver tissue engineering.^{[7a, 62b, 62c, 66]} The lobule is a millimeter-size hexagonal structure consisting of plates of hepatocytes, endothelial cells, biliary epithelial cells, portal fibroblasts, and other supporting cells.^[67] Using digital light processing (DLP) printing technology (Figure 2H), a hydrogel system of human iPSC-derived hepatic progenitor cells, human umbilical vein endothelial cells (HUVECs), and adipose-derived stem cells was assembled using a scaffold of gelatin methacrylate (GelMA) and glycidyl methacrylate-hyaluronic acid (GMHA).^[7a] A two-step printing process was used to layer HLCs and supporting cells to construct a lobule-like pattern. The resulting lobule model measured 3 mm in width and 200 μm in thickness and possessed a hexagonal architecture similar to lobules in vivo, with improved morphological organization and enhanced hepatic functions.

4. Engineering the Microenvironment: Cell-Cell Interactions

Many non-parenchymal cell types in the liver support the functions of hepatocytes, maintain liver homeostasis, and coordinate cell responses to injury and disease progression. On the molecular level, these cells are responsible for secreting growth factors and ECM proteins, thereby creating a microenvironment that facilitates cell-cell and cell-matrix interactions. In this section, we will discuss strategies for recreating the appropriate cell composition and signaling to optimize the microenvironment within liver microtissues (Figure 3A), and highlight examples that illustrate the importance of each non-parenchymal cell type (Figure 3B-F).

Figure 3.

Cell–cell interactions in the hepatic microenvironment. A) Key players include hepatocytes, LSECs, HSCs, and KCs, whose crosstalk in response to liver injury is important for maintaining homeostasis and promoting regeneration. B) A nascent vascular network was formed in a liver bud by co-culturing hepatic progenitor cells with HUVECs and MSCs. C) Transplantation of the liver bud into mice resulted in the formation of mature vasculature, allowing the perfusion of dextran. D) Staining of ECM proteins in rat hepatocyte monoculture and co-culture with HSCs. Picrosirius red and silver impregnation were used to stain collagen and reticulin, respectively. A capsule of fibronectin was identified in the co-culture. E) 3D InSight co-culture model of primary human hepatocytes and KCs showed greater sensitivity to APAP-induced hepatotoxicity when the microtissues were treated with LPS. F) Pro and antiinflammatory interleukin-related genes across various treatments. (B,C) Reproduced with permission.^[42] Copyright 2013, Springer Nature. (D) Reproduced with permission.^[84] Copyright 2005, Karger. (E,F) Reproduced with permission.^[91] Copyright 2019, Elsevier. Parts of this figure were created using Biorender.com.

4.1. Liver Sinusoidal Endothelial Cells

Liver sinusoidal endothelial cells (LSECs) are specialized endothelial cells that make up 15–20% of the cells in the liver.^[68] In contrast to other capillary endothelial cells, LSEC membranes have pores or fenestrae, and lack a diaphragm and basement membrane.^[69] These features allow efficient bidirectional transport of macromolecules across the sinusoidal endothelium.^[70] One side of the endothelium faces the oxygenated arterial blood and portal blood from the gut and pancreas; the other side faces the space of Disse, where LSECs directly come in contact with hepatocytes and hepatic stellate cells (HSCs), which are kept quiescent by LSECs to inhibit their vasoconstrictive effects and maintain a low portal pressure.^[68] LSEC’s fenestrated membrane, transcytosis and scavenging activities confer a selective barrier at the sinusoid, allowing exchange of nutrients, bile acids, and hormones from the blood, and protein, lipid, and glucose metabolic products from hepatocytes.^[68] These characteristics also allow LSECs to mediate immune and inflammatory responses.^[71] In addition, during liver regeneration, LSECs help establish functional tissue architecture by hepatocyte-sinusoid alignment, in which hepatocytes align along the closest sinusoid.^[72]

LSECs regulate hepatocyte phenotype and proliferation via paracrine signaling involving hepatocyte growth factor (HGF) and Wnt2, which help initiate regeneration during injury.^[73] Crosstalk between LSECs and HSCs is important for HSC activation following liver injury. HSCs are activated by paracrine-mediated signals involving fibronectin EIIIA, platelet-derived growth factor (PDGF), and exosomes containing sphingosine kinase 1 (SK1) from LSECs, resulting in the release of cytokines which in turn, alter LSEC gene expression.^[74] Like other endothelial cells, LSECs produce nitric oxide (NO), prostacyclin (PGI₂), and carbon monoxide (CO), which maintain vascular homeostasis and provide protection during liver injury, and endothelin-1 (ET-1), a vasoconstrictor that acts as a counterpart to the vasodilator NO.^{[70, 75]} ET-1 contributes to vascular tone but it is often implicated in endothelial dysfunctions.^[76]

Given the essential role they play in modulating liver function and regeneration, endothelial cells are frequently co-cultured with hepatocytes to enhance the model. LSECs lack a unique surface marker, making their identification and isolation a challenge;^[68] therefore, endothelial cells from other sources have been used to create liver microtissues, including HUVECs,^{[7a, 7b, 7d, 77]} human adipose microvascular ECs (HAMEC),^[78] and cord blood-derived endothelial progenitor cells (EPCs).^[8b] Takebe et al. created a nascent vascular network in liver buds by using HUVECs stabilized by mesenchymal stem cells (Figure 3B).^[7b] Transplantation of liver buds into a preformed cranial window in mice allowed further maturation of the vasculature, which connected to host vessels to allow perfusion of dextran (Figure 3C). Vascularization led to better engraftment and enhanced hepatocyte functions. This research highlights the importance of endothelial cells in the hepatic microenvironment, where they establish a vascular niche with paracrine signaling that upregulates pathways such as FGF and BMP in hepatocytes to sustain their survival. Mimicking organogenesis and introducing endothelial cells at the appropriate time is thus important in creating tissues resembling adult liver.

4.2. Hepatic Stellate Cells

HSCs (HSCs) reside in the space of Disse, between the sinusoidal endothelium and the parenchyma.^[79] Under healthy conditions, HSCs are quiescent, representing 5–8% of the cells in the liver.^[80] They serve as the body’s major storage site for vitamin A, and express markers typical of adipocytes.^[81] Upon liver injury, HSCs undergo a dramatic change in phenotype, transdifferentiating into myofibroblast-like cells and acquiring contractile, proinflammatory, and fibrogenic functions. Activated HSCs proliferate and migrate to the injury site and interact with other cell types to limit injury and repair damaged tissue via ECM remodeling. ^[82] Their contractile properties also impart increased resistance to blood flow in the sinusoid.^[81]

In chronic and acute liver damage, HSC activation is regulated by paracrine and autocrine signals from the microenvironment. Apoptotic hepatocytes release DNA and other molecules that are detected by HSC toll-like receptors (TLRs), initiating a signaling cascade that results in increased synthesis of collagen type I and α-smooth muscle actin (α-SMA), as well as producing a cocktail of chemokines and cytokines including connective tissue growth factor (CTGF), transforming growth factor β (TGF-β), interleukin-6 (IL-6), macrophage chemoattractant protein 1 (Mcp-1), CCL5, and CXCL1. CTGF and TGF-β are profibrogenic and are therapeutic targets for chronic liver fibrosis.^[83] Mcp-1, CCL5, and CXCL1 promote leukocyte recruitment to injury sites,^[79] contributing to an inflammatory and fibrogenic environment that stimulates liver regeneration. In cases where liver regeneration fails to limit injuries, these positive feedback loops result in a vicious cycle of inflammation, leading to fibrosis, cirrhosis, and cancer.^[81]

Since HSCs are responsible for maintaining homeostasis in the hepatic sinusoid,^[80] they are often incorporated in liver microtissues for hepatocyte aggregation and fibrosis studies. HSCs were shown to expedite aggregation of primary rat hepatocytes via self-assembly and in a porous silk scaffold.^[84] In one co-culture model, picrosirius red staining and silver impregnation revealed elaborate matrices of collagen and reticulin, respectively, while immunohistochemistry of fibronectin showed a capsule on the periphery of the spheroid (Figure 3D).^[84a] These structures were absent in the corresponding hepatocyte monoculture, which also took longer to aggregate. In addition to providing ECM support, HSCs secrete HGF and TGF-β to mediate cell injury responses similar to liver regeneration. HSC co-culture with hepatocytes displayed increased hepatic gene expression such as CYP enzymes, albumin, and hepatocyte nuclear factor 4 α (HNF4α), as well as improved functional outputs, including albumin and urea secretions, and metabolism of testosterone and phenacetin.^[84b] In addition, liver cancer microtissues containing HSCs exhibited greater proliferation, epithelial-mesenchymal transition (EMT) potential, and drug resistance than 2D and 3D monocultures.^[85] Taken together, these studies demonstrate the importance of including HSCs in disease models of liver injuries that can lead to fibrosis and cancer.

4.3. Kupffer Cells

Kupffer cells (KCs) are the resident macrophages of the liver, localized in the sinusoid along with other immune cells such as circulating macrophages, neutrophils, natural killer cells, and T cells.^{[79, 86]} KCs are the liver’s first line of defense against pathogens and harmful substances, and interact with other immune cells to maintain the health and functions of the parenchymal tissue. In healthy livers, KCs exhibit a tolerogenic phenotype, clearing toxins from the gastrointestinal tract without inducing inflammatory reactions.^[87] In diseased livers, KCs orchestrate the immune response by displaying a proinflammatory M1 phenotype or an anti-inflammatory M2 phenotype, depending on the metabolic and immune microenvironment.^[87]

Hepatocellular necrosis following injury activates KCs, causing them to undergo a phenotypic change. KCs that adopt an M1 phenotype secrete high levels of cytokines including tumor necrosis factor α (TNF-α), interferon γ (IFN-γ), IL-1, IL-6, IL-12, IL-23, as well as reactive oxygen species (ROS). KCs that adopt an M2 phenotype show reduced production of proinflammatory cytokines such as IL-12, and increased production of anti-inflammatory molecules such as IL-10, TGF-β, PDGF, and ECM components. IL-10 has a protective role during hepatocyte injury.^[88] Activated KCs recruit and activate non-resident immune cells such as macrophages and natural killer cells, and mediate the local immune response. KCs also induce different cell responses from non-immune cells in the microenvironment. LSECs and KCs make up a powerful scavenging system, and the proinflammatory cytokines released by KCs further induce phenotypic changes in LSECs, leading to leukocyte recruitment, capillarization, and formation of a basement membrane (BM),^[71] which in turn activates HSCs.^[68] PDGF produced by KCs, LSECs, and platelets is another potent HSC activator.^[89] KC-secreted TGF-β recruits bone marrow-derived fibrocytes to the liver, where they become activated and differentiate into myofibroblasts.^[90]

Due to the ability of KCs to secrete pro- and anti-inflammatory soluble factors to modulate inflammation, KCs are a key cell type in drug toxicity and immune response studies. Primary human KCs have been used in co-culture with PHHs to study hepatotoxicity using 3D InSight™, a commercial platform for scaffold-free microtissue generation.^[91] Jiang et al. used lipopolysaccharides (LPS), a common bacterial endotoxin, to induce inflammation.^[91a] Microarray and hierarchical cluster analysis revealed that LPS treatment upregulated Fc fragment receptors (FcγRs) and TLR4, activating KCs for phagocytosis and secretion of pro- and anti- inflammatory cytokines. Acetaminophen (APAP), a hepatotoxin, was shown to cause severe mitochondrial damage and reduce phagocytosis through FcγRs when co-administered with LPS. The authors hypothesized that overactivation of TLR4 with LPS on KCs and hepatocytes could exacerbate local inflammation and sensitize the cells to APAP (Figure 3E, F).

4.4. Engineering Complex Microtissues with Multiple Cell Types

3D microtissues composed of hepatocytes, LSECs, HSCs, and KCs possess complex cell-cell contacts and provide more physiologically relevant models for studying disease progression.^[92] Two or more of these non-parenchymal cells have been incorporated into self-assembled liver microtissues in several studies.^{[7c, 93]} Baze et al. generated primary human hepatocyte spheroids in co-culture with LSECs, HSCs and KCs using the 3D InSight™ platform.^[7c] Gene expression and immunostaining confirmed the integration and maintenance of these cell types through 14 days of culture. Upon LPS treatment, IL-6 secretion due to KC activation was significantly greater than in the hepatocyte monoculture. LPS also sensitized the microtissues to the hepatotoxin trovafloxacin. In another study, microtissues were used to model non-alcoholic steatohepatitis (NASH),^[93c] and were treated with palmitic acid for 9 days, which induced proinflammatory and profibrogenic responses, as well as apoptosis in the microtissues. Activation of HSCs resulted in the production of collagen, α-SMA, tissue inhibitor of matrix metalloproteinase-1 (TIMP-1), PDGF receptor β, and plasminogen activator inhibitor-1 (PAI-1, a TGF-β target protein). Upregulation of IL-8 was indicative of KC activation and an inflammatory microenvironment.

Three-dimensional liver microtissue models incorporating multiple cell types are attractive due to their physiological relevance, but it remains challenging to select the appropriate media and ratio of cell types. Hepatocytes, LSECs, HSCs, and KCs are sensitive cell types that require specific culture conditions. Despite the extraordinary self-organizing behavior of these cells, it is often difficult to replicate specialized hepatic architectures such as microvasculature, the space of Disse, and hepatic lobules within microtissues. Biomaterial and culture system optimization are necessary to improve the 3D models, and will be discussed in the next two sections.

5. Engineering the Microenvironment: Cell-Matrix Interactions

A range of natural, synthetic, and hybrid polymers with varying chemical compositions and mechanical properties have been developed to support cell-cell and cell-matrix interactions and 3D liver tissue formation. These polymers can improve in vitro hepatocyte culture and disease models versus traditional 2D and some self-assembly methods. In this section, we discuss the design of biomimetic biomaterials and recent advances in natural and synthetic scaffold development.

5.1. Designing Biomimetic Scaffolds

The extracellular matrix is an important component of the hepatic microenvironment, providing structural support and adhesion for resident cells. The ECM is highly dynamic and regulates cell behavior by changing its composition during tissue development, regeneration, and disease progression. The ECM regulates the hepatic microenvironment by binding cell surface receptors and inducing intracellular signaling, as well as by binding and releasing growth factors, thereby altering local growth factor concentration gradients.^[94] When selecting biomaterials for use as ECM-mimetic scaffolds for liver microtissues, it is important to consider their chemical, mechanical, and physical properties (Figure 4A).

Figure 4.

Cell–matrix interactions in the hepatic microenvironment. A) Some parameters for consideration in the design of biomimetic scaffolds for liver tissue engineering. B) Proteomic analysis of ECM composition after decellularization of porcine liver. C) Murine iHep/HUVEC cultured in a 3D hydrogel of decellularized ECM under perfusion (iHE-F) demonstrated superior activities versus static monoculture (iH-S), co-culture (iHE-S), and perfused monoculture (iH-F). D) Human iHep on synthetic scaffolds of varying PMCL:PCL ratios; a 50:50 ratio showed the lowest Young’s modulus and greatest albumin secretion. E) Hydrogel of heparin and PEG incorporated HGF, which was slowly released over time. F) Albumin and urea secretion were better maintained in the hybrid hydrogel containing HGF. G) Primary human hepatocytes cultured on electrospun PLGA scaffolds linked with collagen (C) or fibronectin (F). (B,C) Reproduced with permission.^[7] Copyright 2018, Wiley. (D) Reproduced with permission.^[115] Copyright 2019, Elsevier. (E,F) Reproduced with permission.^[117] Copyright 2010, Elsevier. (G) Reproduced with permission.^[117] Copyright 2018, Elsevier. Parts of this figure are reproduced under the terms of the Creative Commons Attribution 3.0 Unported License, Copyright Servier Medical Art by Servier.

The ECM presents a variety of biochemical cues that guide cell movement, proliferation, morphology, and differentiation.^[95] Integrins on the plasma membrane bind specific motifs of ECM proteins, forming macromolecules called focal adhesions, through which extracellular signals are sensed. The intracellular domains of integrins are bound to other signaling proteins and the cytoskeleton through adaptor proteins, relaying signals sensed at the focal adhesions to the cell body. The major constituents of hepatic ECM include collagen types I, III, IV, V, and VI, fibronectin, laminin, and proteoglycans.^[96] Liver microtissue scaffolds should include bioactive residues derived from these ECM components that facilitate cell-cell and cell-matrix interactions, to provide proper signals for cell aggregation and orientation.

Cell behavior in the hepatic microenvironment is also regulated by mechanical cues. The liver has a stiffness between 2.3 and 5.9 kPa,^[97] thus biomaterials with a low Young’s Modulus (E) are desired in liver tissue engineering. Differentiation of bipotent liver progenitor cells integrates both chemical and mechanical cues, where hepatocyte and biliary cell fates correlate with reduced and increased substrate stiffness, respectively.^[98] By constructing polyacrylamide (PA) hydrogels of high (E = 30 kPa) and low (E = 4 kPa) stiffness followed by collagen type I and Notch ligand patterning, both hepatocytes and biliary cells can be differentiated from bipotent progenitors. High substrate stiffness increased Notch signaling and biliary differentiation at the periphery, whereas low substrate stiffness favored HNF4α expression and hepatocyte differentiation in the center. LSECs are also sensitive to mechanical perturbations such as shear stress or stretching, and release NO and ET-1 to modulate vasodilation and constriction.^{[95, 99]} Mechanotransduction alone is sufficient to turn on LSEC angiocrine signals that support the survival and proliferation of isolated PHHs.^[100] This finding can be translated to liver tissue engineering by using a perfusion-enhanced system to facilitate mechanosensing at the cell-matrix interface.

5.2. Natural Scaffolds

Natural polymers are attractive for tissue engineering due to their biocompatibility and bioactive properties. Natural polymers used as scaffolds in liver microtissues include collagen,^[49] heparin,^[101] hyaluronic acid (HA),^[102] fibrin,^[103] Matrigel,^[104] alginate,^{[8b, 105]} chitosan,^[106] and silk.^[107] Since >60% of human liver ECM consists of various types of collagen, they are used most often to provide structural and mechanical support in microtissues.^[108] As glycosaminoglycans (GAGs), heparin and HA are popular in preserving the bioactivity of growth factors in forming hydrogels.^[109] Alginate and chitosan are non-mammalian polysaccharides, but they are attractive for tissue engineering due to their biocompatibility, biodegradability, and low cost.^[110] Alginate is commonly used for cell encapsulation due to its low cell adhesiveness, which is helpful in promoting cell-cell interactions and spheroid formation.^[111] Moreover, alginate encapsulation protects microtissues from immune cells and physical damage, making it suitable as a cell therapy delivery system.

Although the individual composition of the mammalian ECM is known, it remains a challenge to artificially engineer the structural and biochemical complexity in vitro. To recreate the native ECM composition, some groups have used decellularized animal liver ECM as a scaffold for microtissue generation.^{[7d, 112]} Jin et al. decellularized porcine livers with detergent followed by enzymatic processing to produce a soluble decellularized liver ECM mixture.^[7d] Almost 80% of the GAGs remained in the mixture, and proteomic analysis revealed the presence of collagens, proteoglycans, glycoproteins, and keratins (Figure 4B). The decellularized liver ECM was used as a 2D substrate or as a 3D hydrogel. Murine iHeps and HUVECs were cultured within the decellularized liver ECM hydrogel under constant perfusion in a microfluidic device. The resulting organoids had a diameter of 2 mm and a height of 1 mm, and showed improved cellular organization and hepatic functions such as albumin secretion, CYP activity, and sensitivity to hepatotoxicity (Figure 4C).

5.3. Synthetic and Hybrid Scaffolds

Natural polymers have inherent variabilities and poor mechanical strengths, whereas synthetic polymers offer greater consistency, mechanical strength, and tunability in the manufacturing process.^[113] A variety of synthetic polymers have been developed for hepatic liver microtissue encapsulation, including poly(ethylene glycol) diacrylate (PEGDA),^[114] polycaprolactone (PCL),^{[41c, 115]} and poly(lactic-co-glycolic acid) (PLGA).^[116] Polymer scaffold porosity, topography, and biocompatibility vary depending on substrate chemistry and manufacturing process (Figure 2), and are important parameters for optimization.^[57b] For example, by varying the weight ratio of amorphous poly(4-methyl-ε-caprolactone) (PMCL) and semi-crystalline poly(ε-caprolactone) (PCL) followed by emulsion freeze-drying, scaffold properties such as pore size, porosity, wettability, and stiffness can be controlled.^[115] By comparing scaffolds of different PMCL:PCL blending ratios, it was found that high wettability and low Young’s Modulus were beneficial for human iHep culture. iHep showed the highest urea synthesis and albumin secretion levels on a scaffold formed using a 50:50 PMCL:PCL mixing ratio, which had an average pore size of 45.9 μm, a porosity of 87.8%, a water uptake ability of 552.5%, and a Young’s Modulus of 29.2 kPa (Figure 4D).

Synthetic polymers exhibit better mechanical properties and tunability than natural polymers, but lack cell recognition signals and their degradation products might be toxic.^{[111, 113]} For this reason, many groups have developed hybrid scaffolds with suitable mechanical strength while providing biocompatibility.^{[105c, 117]} Kim et al. cultured primary rat hepatocytes in a hydrogel of thiolated heparin and acrylated poly(ethylene glycol) (PEG).^[117b] Due to heparin’s ability to bind growth factors, the scaffold incorporated HGF and released it slowly over time to support hepatocytes, a process similar to ECM degradation and HGF release in vivo. The resulting 3D spheroids were 236 μm in diameter and functional after 20 days of in vitro culture (Figure 4E,F). In another study, an optimized technique was used to wet electrospin PLGA to create a network with an average pore size of 27.5 μm, larger than the average pore size generated by conventional electrospinning, and similar in size to hepatocytes.^[117c] Collagen type I and fibronectin were immobilized onto PLGA before PHHs were seeded and allowed to penetrate the 3D scaffold (Figure 4G).

6. Engineering the Culture System

Perfusion-based cell culture systems offer the advantages of mechanical stimulation, constant renewal of culture medium, and intercellular signaling. In this section, we examine two perfusion systems for hepatic culture: bioreactors and microfluidic chips. Bioreactors are useful in large-scale manufacturing and have great potential for translational medicine. Microfluidic systems allow fine control and manipulation of biological systems. Both are superior to static cultures, but challenges remain in increasing their adaptability and reducing their costs of operation.

6.1. Bioreactors

Bioreactors recapitulate the native environment of tissues and facilitate mass transport via a continuous supply of nutrients and oxygen, as well as the removal of waste metabolites. Established bioreactor designs allow control of environmental parameters, such as oxygen level, pH, nutrient supply, shear stress, and extracellular matrix composition. Compared to static cultures, bioreactors support tissue constructs with prolonged viability and more mature liver-specific functions, such as synthesis of albumin and CYP enzyme.^{[102a, 118]} Oxygen concentration, a major factor investigated in bioreactor design, regulates the metabolic zonation of liver via differential spatial expression of genes encoding carbohydrate-metabolizing enzymes and oxygen-responsive transcription factors.^{[63a, 119]} Shear stress, another frequently examined factor, affects the rate of albumin and urea synthesis by hepatocytes.^[120] The shear stress in normal human hepatic sinusoids is less than 5 dynes/cm².^{[120b, 120d]} Studies using microfluidic chips showed that low shear stress (0.01–0.33 dynes/cm²) increased cell viability and hepatic function than high shear stress (5–21 dynes/cm²),^{[120a, 120b, 121]} thus volumetric flow rates within the culture system can be adjusted within the physiologically relevant range to optimize flow profiles.

Here, we highlight recent advances in bioreactor design that prolong hepatocyte viability and function. Hollow fiber membrane bioreactors provide an interstitial space to maintain liver spheroids for up to 25 days and allow variable parametric tests including urea synthesis, albumin production, and biotransformation (Figure 5A).^{[119e, 122]} Previously, hollow fiber bioreactors used hydrogels to encapsulate cell suspensions in the lumen, allowing their aggregation over time and perfusion upon gel shrinkage.^[123] In more recent studies, hepatocyte spheroids were first formed in gel droplets prior to seeding on the outer surface of microporous membranes.^{[119e, 122, 124]} These methods showed higher throughput and robustness, and allowed more adjustable input parameters such as multi-axis flow profile, spheroid diameter, inter-hollow-fiber spacing distance, and membrane porosity. Bioreactors can also control hepatocyte spheroid formation in a gel-free manner via an acoustofluidic device, as demonstrated in one study using acoustic waves to precisely control the position and size of aggregates of human hepatoma Huh7 cells.^[125]

Figure 5.

Engineering optimization of culture systems. A) Hollow fiber bioreactor allows long-term culture of liver microtissues, by facilitating the exchange of nutrients, oxygen and waste products. Spheroids were first encapsulated in hydrogel before dispersing on the hollow fiber surface. B) Bioreactor as part of a bioartificial liver system. C) Schematics of liver-on-chip models: i) liver spheroids on a chip; ii) self-assembled liver microtissues on a chip; iii) sinusoid-on-chip, which consists of a layer of porous membrane sandwiched by ECs and HSCs/hepatocytes. Panel B: reproduced with permission. ^[126] Copyright 2016. Springer Nature.

Large-scale bioreactors are essential in bioartificial liver devices (BALs, more details in Section 7), extracorporeal systems that assist patients with diseased livers (Figure 5B).^[126] In one BAL study, three billion human iHeps were maintained in a multi-layer radial-flow bioreactor consisted of 65 layers of polycarbonate plates, which promoted cell adhesion and hepatic gene expression.^[126b] When the bioartificial liver was assisting pigs with acute liver failure, it rescued 7 of 8 pigs by providing metabolic detoxification of ammonia and bilirubin, mitigating liver damage by reducing alanine aminotransferase (ALT) and aspartate aminotransferase (AST) levels, and synthesizing albumins, for over 7 days.

6.2. Microfluidic Systems

In static cultures and in bioreactors, it is challenging to achieve precise control of microenvironment parameters such as chemical gradients, shear stress, and cell patterning. Organ-on-chip technologies provide a solution to this problem.^[127] Due to the small scale of microfluidic channels (<1 mm), fluid flow is laminar, enabling the generation of physical and chemical gradients. Complex structural features such as porous membrane,^[128] microposts,^[129] and electrical/mechanical devices can be incorporated in microfluidic systems to better recapitulate and control structural and functional features of living tissues.^{[128c, 130]} Liver-on-chip systems using 3D microtissues can potentially be used to recreate an architecture similar to that of sinusoids in vivo (Figure 3A).

One advantage of liver-on-chip systems is controlled perfusion, which helps maintain liver spheroids embedded in biomimetic scaffolds (Figure 5C, i).^{[119a, 129f, 131]} Continuous flow allows transport of nutrients and oxygen, as well as the removal of metabolic waste. To avoid flow imparting high shear stress to hepatocytes, Lee et al. used closely spaced microposts to separate cell culture chambers from perfusion channels.^[129f] With this design, hepatocytes were cultured in a low shear environment while rapid mass transport was sustained by diffusion, resembling the fluidic niche of the hepatic sinusoid.

Liver-on-chip systems can be used to co-culture hepatocytes with non-parenchymal cells. Vernetti et al. created a sequentially layered, self-assembly liver (SQL-SAL) model that included PHHs, endothelial cells, immune cells, and stellate cells, which self-organized into a layered structure (Figure 5C, ii).^[132] The resulting system formed a favorable microenvironment for hepatocytes, promoting albumin and urea secretions, and supporting cell survival for over 28 days. To reproduce the space of Disse, an unique interface between hepatocytes and the endothelium, a porous membrane could be used to separate the two cell types.^{[105c, 133]} Prodanov et al. improved this model by embedding HSCs in a collagen gel between hepatocytes and the endothelium, creating the space of Disse for HSCs (Figure 5C, iii).^[134]

Liver-on-chip systems can be coupled with other organotypic cultures such as cardiomyocytes, islets, gastric organoids and tumor tissues. These integrated microphysiological systems (MPS) mimic organ-organ interactions, and facilitate the evaluation of metabolism, efficacy, and toxicity of pharmaceuticals in a systematic manner. In a closed perfusion system, circulating drugs are metabolized in the liver chamber and their metabolic products are delivered to other tissues, where they might have therapeutic or side effects.^[135] MPS consisted of the appropriate tissue composition approximates the human organ systems and can bridge the gap between traditional cell culture and animal models.

7. Applications and Challenges of Liver Microtissues for Disease Modeling and Regenerative Medicine

Liver microtissue models have been developed for biomedical applications such as organogenesis, disease modeling, drug testing, hepatotoxicity screening, and regenerative medicine (Figure 6). Following the theme of this special issue, we discuss applications and challenges of 3D liver microtissues for disease modeling and regenerative medicine.

Figure 6.

Applications of engineered liver microtissues. Disease modeling and regenerative medicine are highlighted in this review. Created using Biorender.com.

7.1. Disease Modeling and Diagnosis

Infectious diseases involving hepatitis B and C viruses,^[136] dengue virus,^[137] and malaria^[138] have been recapitulated in vitro. Cellulosic sponges have been used to culture human and simian primary hepatocyte spheroids, and the long-term cell viability and hepatic characteristics permitted malaria parasites to fully develop from sporozoites.^[138] These hepatocyte spheroids also provided a platform to screen drugs for malaria treatment. HLCs derived from specific patients’ iPSCs are a powerful tool in large-scale genomic studies^[139] and in investigating molecular pathways of genetic diseases such as α−1-antitrypsin deficiency,^[140] familial hypercholesterolemia,^[141] abetalipoproteinemia,^[142] and Wolman Disease.^[43] Multi-cellular human liver organoids incorporating HLCs, KCs, and HSCs were differentiated from 11 human iPSC lines, some of which were from patients with Wolman Disease.^[43] The patient organoids demonstrated progressive steatohepatitis observed by lipid accumulation and increased stiffness, indicating the onset of fibrosis.

Challenges remain in using liver microtissues in disease modeling. The advantages and disadvantages of the starting hepatocyte source, including primary cells, cell lines, HLCs, and iHeps (Figure 1), must be evaluated carefully as they may impact the validity and reproducibility of the liver model. Additional improvements such as the maturity of HLCs and iHeps, cell signaling dynamics, and ECM properties during disease progression, are needed in the continuous optimization of 3D microtissues. Most diseases affect multiple tissue types; therefore, MPS that integrates multiple organs and permits their crosstalk can provide valuable insights into disease pathways. A hepato-biliary-pancreatic organoid model was recently developed to study human endoderm organogenesis and pathology.^[44] When an embryonically lethal genetic defect in the transcription factor Hes1 was created in iPSCs, biliary specification was abolished in the integrated organoid. Although hepatocytes developed normally, the other liver parenchymal tissue (consists of biliary cells) could not, thus hindering bile transport. Lastly, the time scale necessary for chronic liver disease initiation and progression is long and difficult to recapitulate in vitro. For example, non-alcoholic fatty liver disease (NAFLD) typically affects middle age adults and can progress slowly to inflammation, cirrhosis, and cancer over the course of many years, depending on the patient’s lifestyle.^[143] Feeding cells with free fatty acids is typically done in NAFLD research, but is an oversimplification of this complex phenomenon.

The development of a large library of human liver microtissues can potentially capture a spectrum of disease phenotypes, enabling large-scale genetic studies, disease diagnosis, and drug screening. This library could be constructed from iPSC banks comprised of diverse genetic and environmental backgrounds, and the resulting liver microtissues must meet preset benchmarks. Such a library coupled with genome editing tools (e.g. CRISPR/Cas9) and high-content analysis, will be a powerful tool for disease modeling and preclinical drug development, advancing our knowledge of human metabolism and drug response.

7.2. Regenerative Medicine

Regenerative medicine focuses on the repair or replacement of damaged tissues and organs to restore or re-establish normal functions.^[144] Engineered liver microtissues are applicable to regenerative medicine through two means: cell therapy and bioartificial liver devices.

7.2.1. Cell Therapy

Direct transplantation of cells into damaged tissue sites offers several advantages: it is less invasive compared to whole organ transplant and allows the damaged organ to regenerate; the cells can be autologous and genetically modified before transplantation.^[145] Clinical trials have focused on cell therapy using primary hepatocytes and these studies have demonstrated the overall safety of the procedure, though the data are scarce. Human fetal liver cells were recently used as well, through intrasplenic injection in patients with end-stage liver disease.^[146] During the one-year follow-up period, cell infusion was safe and well tolerated, with reduction of cirrhosis and disease severity. Encapsulation of hepatocytes (e.g. in alginate) can provide anchorage in the tissue and protect the cells from immune rejection, and potentially enhance cell-cell and cell-matrix interactions to prolong cell survival in vivo.^[147] In a recent study, human HLC spheroids were shown to rescue mice with acute liver injury.^[41c] Intraperitoneal transplantation of HLC spheroids reduced damage from partial hepatectomy, and the mice gained weight starting 10 days earlier than the untreated group.

Many challenges remain before liver cell therapy is ready for patients. Hepatocyte engraftment has low efficiency in vivo, as in other types of cell transplantation. Several approaches have been used to improve liver cell engraftment including reversible portal vein embolization (RPVE)^[148] and hepatic irradiation.^[149] However, the biggest challenge is obtaining an expandable cell source and a large number of cells. Intraportal infusion typically requires 10⁸ cells/kg,^[147a] requiring large-scale manufacturing with excellent quality control. This problem could be addressed by harnessing the proliferation potential of iPSCs, to obtain enough cells from a single source. In addition, instead of lifelong reliance on immunosuppression in traditional organ and cell therapies, a library of human leukocyte antigen (HLA) homozygous iPSCs could be used to obtain HLCs that match each patient’s HLA.^[150] However, other optimization is needed before iPSC-derived products are ready for clinical use, including safety, cell maturity, and product purity.

7.2.2. Bioartificial Liver Devices

Due to the limited availability of healthy donor organs, liver and hepatocyte transplants are currently not possible for every patient. Complications from patient management to surgical techniques also result in variable clinical outcomes. An alternative to transplantation is the use of extracorporeal bioartificial liver devices (BALs). BALs are external bioreactors that contain hepatocytes that satisfy the minimum liver functional requirements of ammonia elimination, drug metabolism, and blood protein synthesis.^[151] Inside these bioreactors, hepatocytes are cultured on one side of a synthetic membrane through which nutrients, oxygen and metabolites are exchanged between the cells and perfusion fluid (Figure 5A).^[152]

Since BAL systems require a large number of cells (approximately 10 billion),^[153] most devices in clinical development employ immortalized human hepatocyte cell lines and primary porcine hepatocytes.^[154] Traditional BALs contain cell suspensions, but BALs consisting of spheroids have also been developed since 3D culture systems have been shown to significantly improve hepatocyte survival and functionality.^{[154c, 155][156]} In a study by Lee et al., isolated porcine hepatocytes were cultured in large spinner flasks with stirring to generate 70 μm spheroids.^[156a] The spheroids were subsequently encapsulated in alginate hydrogel and cultured in a BAL bioreactor to recirculate and detoxify blood. In a porcine acute liver failure model, the spheroid-based BAL demonstrated effective ammonia clearance and renal function preservation.

As in hepatocyte transplantation, BALs face the challenge of finding a suitable cell source, which results in their limited clinical efficacy.^[153] Due to the many metabolic functions that hepatocytes perform, it is difficult to standardize a set of predictive markers to measure and compare the functions of all existing bioartificial liver devices. However, for the successful large-scale manufacturing and commercialization of BALs, it would be invaluable to establish specific standard markers to evaluate existing devices before use in patients. Step-by-step optimization of the microenvironment and culture system, as discussed above, is applicable to the systematic development of BALs, which have great potential as a temporary bridge before patients receive whole liver transplants or cell therapy.

8. Conclusions and Perspectives

Tremendous progress has been made in the field of liver tissue engineering. Liver microtissues have emerged as a superior system for disease modeling and regenerative medicine. This review discusses their fabrication strategies, and optimizations of the hepatic microenvironment and culture system for creating microtissues that are physiologically relevant liver models and pave the way to regenerative medicine. With advances in cellular and molecular engineering, hepatocyte cell sources have demonstrated improved in vitro survival and functionality over prolonged periods of time. Liver microtissues can be constructed via self-assembly of mature hepatocytes or organoids, or via scaffold-assisted methods. Incorporation of non-parenchymal liver cells enhances cell-cell interactions and promotes secretion of niche-specific factors important to liver model accuracy. Incorporation of natural, synthetic, or hybrid polymeric materials offers controllable and tunable properties for building biomimetic scaffolds that enhance cell-matrix adhesion, modulate cell behaviors, and provide immune and physical protection. Perfusion-based culture systems allow greater control and manipulation of the biological system to recreate the complex environment that hepatocytes experience in vivo.

The structural and functional complexities of the native liver pose intrinsic challenges for engineering liver microtissues.^[157] Unlike many other organs which carry out a relatively low number of specialized tasks, liver performs over 500 synthetic and metabolic functions.^[158] It is difficult to set benchmarks against these many functional requirements, and engineered liver models must be carefully evaluated for each application. Although the co-culture of hepatocytes and non-parenchymal cells can potentially improve the physiological relevance of liver microtissues, they are all extremely sensitive cell types that will undergo drastic phenotypic and functional changes under stress, resulting in observed donor-to-donor and batch-to-batch variabilities in experimental outcomes. Understanding the supportive roles of the non-parenchymal liver cells such as LSECs, HSCs and KCs will be important for optimizing co-culture ratio, medium composition, and culture conditions. Recent studies of engineered ECM scaffolds have provided certain cues to guide cellular organization and behaviors, but hepatic architectures such as microvasculature and lobule remain difficult to recapitulate within microtissues. Systematic optimization of biomimetic materials in concert with tissue-chip designs will further aid the development and maturation of delicate structures in liver models.

3D liver microtissues are unique engineered tissue constructs that have proven to be superior than traditional monolayer hepatocyte cultures, insightful disease models, and potential cellular systems for precision medicine (applications to organogenesis and pharmaceutical testing are reviewed elsewhere). Many challenges remain before they can be used in large-scale manufacturing and clinical applications, as discussed in Section 7. As liver metabolic activities vary from species to species, finding an expandable and reliable source of human cells is one of the most pressing issues in biomedicine. Human iPSCs can potentially address this, and they have been differentiated into homogeneous HLCs and complex organoids. Multi-cellular organoid models are highly structured and superior for studying organogenesis and diseases, and for drug and hepatotoxicity screening. However, their higher heterogeneity and variability are undesirable for clinical applications. For this purpose, HLCs are attractive because they can be produced in large batches and analyzed to determine their suitability for therapeutic use, before purification and 3D encapsulation for cell therapy or bioartificial liver devices. Integral to moving tissue engineering products towards commercial and clinical applications is biomanufacturing.^[159] Current differentiation and culture of hepatocytes rely on growth factors and natural scaffolds, which will need to pass standardized quality controls. Isolation, purification and manipulation of cells, as well as their packaging, delivery and preservation, are all important considerations for the biomanufacturing pipeline.^[160] Material engineering, bioreactor design and automation can further support the large and consistent production of hepatic cells.^[159]

Despite its challenges, liver tissue engineering has come a long way, and along with advances in genome engineering and next-generation sequencing, may improve the treatment of hepatic diseases via disease modeling, precision diagnosis, and drug screening. Moreover, engineered microtissues can someday be delivered to the injured liver for tissue regeneration or used for whole organ engineering, alleviating the shortage and complications of liver transplants.

Author Biographies and Photographs

Dantong (Danielle) Huang received her B.S. and M.Eng. in Biological/Biomedical Engineering at Cornell University, before joining the Ph.D. program in the Department of Biomedical Engineering at Columbia University. She is currently in Dr. Kam Leong’s laboratory and working on generating donor-specific liver microtissues for disease modeling and therapeutic applications.

Sarah B. Gibeley received her B.S. in Nutritional Science at Boston University in 2010 and served as a research technician at the Tufts University Human Nutrition Research Center on Aging. She is a Ph.D. candidate in the Institute of Human Nutrition at Columbia University and is currently working on investigating the genetic causes of non-alcoholic fatty liver disease in the laboratory of Dr. Henry N. Ginsberg.

Kam W. Leong is the Samuel Y. Sheng Professor of Biomedical Engineering at Columbia University, with a joint appointment in the Department of Systems Biology at Columbia University Medical Center. His research focuses on nanoparticle-mediated drug-, gene-, and immunotherapy, from the design and synthesis of new carriers to applications for cancer, hemophilia, infectious diseases, and cellular reprogramming.