HUMAN iPSC-DERIVED MULTICELLULAR LIVER MODEL FOR HIGH-THROUGHPUT DRUG SCREENING

Abstract

3D bioprinting technology is used to reconstruct a multicellular liver acinus through precise positioning of multiple human iPSC-derived cell types, including immune cells, in a configuration that mimics the physiological cell-cell interactions in a native liver microenvironment. The liver model recapitulates hepatic organization by spatially positioning endothelial and parenchymal compartments in physiologically relevant arrangements, enabling paracrine signaling, enhanced nutrient exchange, and stable cell-cell / matrix interactions. When integrated with a dynamic microfluidic perfusion system, the liver model supports matrix remodeling, delivers physiological shear stress, and sustains a well-oxygenated microenvironment.

Description

HUMAN iPSC-DERIVED MULTICELLULAR LIVER MODEL FOR HIGH-THROUGHPUT DRUG SCREENINGRELATED APPLICATIONS

[0001] This application claims the benefit of the priority of U.S. Provisional Application No. 63 / 753,410, filed February 3, 2025, which is incorporated herein by reference in its entirety.FIELD OF THE INVENTION

[0002] The present invention relates to a platform for high-throughput drug screening through replication of the liver’s complex architecture and functionality in vitro via an iPSC-derived multicellular liver model. By incorporating patient-specific iPSCs, the inventive approach supports personalized drug testing.BACKGROUND

[0003] Animal models have long been central to preclinical drug development, yet their ability to predict human hepatotoxicity is limited. In fact, animal studies forecast only -55% of liver-related adverse drug reactions observed in humans, underscoring a major translational gap. This poor predictivity arises from interspecies differences in liver enzyme expression, metabolic pathways, and immune responses — factors that frequently result in unanticipated toxicity during clinical trials and contribute to high drug attrition rates. Recognizing these limitations, regulatory agencies are beginning to shift toward more human-relevant testing strategies. In 2024, the U.S. Food and Drug Administration (FDA) announced plans to phase out mandatory animal testing for biologies, including monoclonal antibodies, marking a major policy shift. Similarly, the National Institutes of Health (NIH) recently issued guidance encouraging the adoption of three-dimensional (3D) in vitro models as alternatives to traditional animal studies. These actions align with the “3Rs” principle — replacement, reduction, and refinement of animal use — and reflect a global movement toward next-generation preclinical tools

[0004] The liver serves a crucial function in synthesizing essential proteins and metabolizing xenobiotics. It follows that liver dysfunction is closely linked to disease development and drug-induced toxicity. Numerous genetic and metabolic disorders can affect the liver, such as fibrosis and cirrhosis. The liver is a critical target organ for drug screening due to its central role in drug metabolism and detoxification, accounting for over75% of systemic drug clearance via enzymatic biotransformation, transporter-mediated uptake and efflux, and biliary excretion. Hepatocytes express a broad repertoire of cytochrome P450 (CYP) enzymes that mediate phase I and II metabolic reactions, influencing drug-drug interactions, prodrug activation, and the formation of toxic metabolites. Given the liver's central role in xenobiotic metabolism, parenchymal cell cultures are indispensable in vitro models for studying detoxification processes. Consequently, extensive efforts have been made to develop in vitro liver models, both for investigation of pathophysiological mechanisms and as accessible alternatives to animal models during drug hepatotoxicity screenings. Primary hepatocyte cultures have emerged as a promising tool for detecting general or liver-specific toxicity and evaluating speciesspecific drug effects. While various approaches for isolating and cultivating primary hepatocytes have been successful, their utility is often limited by challenges in maintaining viability and metabolic function.

[0005] Liver cancer is a leading cause of cancer-related deaths, particularly among populations exposed to environmental toxins such as chemicals, pollutants, and carcinogens. These substances can disrupt the liver's immune regulatory functions and lead to cancer development. Military personnel, who are frequently exposed to such harmful substances during deployments or occupational duties, face a heightened risk of liver cancer. Existing liver models fail to accurately replicate the liver’s complex immune physiology and multicellular architecture, limiting their ability to predict the carcinogenic potential of environmental exposures.

[0006] Conventional in vitro liver models — typically 2D monocultures — lack the architectural complexity, multicellular interactions, and dynamic microenvironments required to maintain hepatocyte phenotype and function. This leads to rapid functional decline, underestimation of hepatotoxic risk, and missed detection of clinically relevant toxicities as well as potential liver cancer risks. In traditional 2D cultures, primary hepatocytes typically cannot survive beyond 2 weeks, quickly losing their differentiated state and metabolic functions. In response, more physiologically relevant liver models are being developed, including 3D spheroids, organoids, and microfluidic “liver-on-chip” systems. Organoids can self-organize into tissue-like structures and reproduce some aspects of liver physiology, making them valuable for disease modeling and certain drug testing applications. However, they exhibit significant variability in size, shape, and cellularcomposition, as well as limited spatial control — resulting in batch-to-batch heterogeneity and functional inconsistency. These limitations hinder their ability to reproduce the highly ordered liver microanatomy essential for coordinated hepatic metabolism and accurate toxicity prediction

[0007] In contrast, in vitro 3D cultures of primary hepatocytes have shown considerable promise in maintaining hepatocyte viability, stable phenotype, morphology, and metabolic functions over extended culture periods. The superior performance of 3D models in preserving hepatocyte-specific functions underscores their importance in studying the biotransformation of xenobiotics. This shift towards utilizing 3D models reflects a broader trend towards more physiologically relevant in vitro systems for drug development and toxicology studies.

[0008] 3D bioprinting has gained considerable attention for its ability to fabricate complex, cell-laden architectures that mimic organotypic structures. Compared to extrusion- or inkjetbased bioprinting approaches, digital light processing (DLP)-based 3D bioprinting - a projection photopolymerization-based technique - offers high spatial resolution, a rapid fabrication speed, and versatile geometric design capabilities, enabling precise spatial control of both cells and matrix components. These capabilities are particularly advantageous for reproducing the intricate microscale and diverse cellular composition of the liver. As a result, DLP bioprinting provides an ideal platform for engineering liver tissue models that closely recapitulate native liver architecture and function.

[0009] Native human tissues typically exhibit a cell density ranging from 1 to 2 billion cells / mL. However, traditional bioink formulations generally have cell densities in the range of 0.1-10 million cells / mL, which is orders of magnitude lower than the levels found in native tissues like the liver. Use of low density bioinks has arisen because higher dispersed cell densities significantly reduce bioink printability and cell viability. Yet, primary hepatocytes thrive in high cell density (HCD) environments, which promote enhanced cellcell and cell-extracellular matrix (ECM) interactions necessary for maintaining cellular function and phenotype. In contrast, these interactions are limited in traditional 2D cultures. HCD is essential for accurately recapitulating various pathological tissue states, such as regeneration and function, in which perturbed cell-cell and cell-ECM interactions play a central role. Therefore, achieving HCD bioinks is crucial for the development of more physiologically relevant tissue models and enhancing understanding of tissue pathology.

[0010] Biofabrication technologies, such as digital light processing (DLP)-based 3D bioprinting, play a critical role in tissue engineering by facilitating the programmed assembly of cells into intricate 3D geometries with high resolution, cell viability, and speed. While extrusion-based printing can achieve a nominal resolution of 50 pm and light-based printing can reach several microns, achieving finer features requires precisely optimizing fabrication conditions, typically by using materials that exhibit low-biocompatibility, high photosensitivity, and poor cell encapsulation. Researchers have developed strategies to create high-resolution bioprinted hepatocyte models, however, even these studies are limited to relatively low cell densities (< 20 million cells / mL). Additionally, the use of cell-laden bioinks in such bioprinting applications can degrade print resolution, leading to reduced accuracy. To increase cell density, there is growing interest in using cellular spheroids to engineer functional hepatic tissues with the requisite HCD. Increased cell density or the use of spheroids in bioink necessitates a larger nozzle tip for extrusion-based 3D bioprinting, which lowers print resolution. Additionally, hepatocytes, being larger cells with a diameter of 20 to 30 pm, face significant shear forces during extrusion, leading to impaired cell viability. Although DLP-based 3D bioprinting is free from extrusion-based shearing concerns, light-based technologies have primarily been designed to process cells embedded within hydrogels, which limits cell-cell interactions and results in low cell density (LCD) constructs. Attempts to increase cell density in hydrogel-based bioinks typically results in severe cell-induced light scattering, causing errant photopolymerization that deforms the print structure. In sum, simultaneously achieving HCD (> 20 million cells / ml), high cell viability (>80%), and finely resolved hepatic tissue constructs remains a significant challenge.SUMMARY

[0011] In vitro liver tissue models are valuable for studying liver function, understanding liver diseases, and screening candidate drugs for toxicity and efficacy. While three-dimensional (3D) bioprinting has shown promise in creating various types of functional tissues, current efforts to engineer a functional liver tissue struggle to replicate native high cell density and ensure long-term cell viability. Achieving high cell density (HCD) enables the cell-cell interactions needed to mimic the liver's metabolic and detoxification functions.

[0012] The inventive scheme leverages advanced bioprinting technology to reconstruct a multicellular liver acinus, i.e., the smallest functional unit of the liver, through precisepositioning of multiple human iPSC-derived cell types, including immune cells, in a configuration that mimics the physiological cell-cell interactions in a native liver microenvironment. Light scattering within the construct is controlled through precise tuning of optical properties of the bioink. In some embodiments, a refractive index matching agent, e.g., iodixanol (IDX) is incorporated into the bioink formulation. This approach enables the fabrication of hepatic tissue constructs with an HCD of 8 x 107cells per mL while maintaining high cell viability (-80%). The printed dense hepatic tissue constructs exhibit enhanced cell-cell interactions, as evidenced by increased expression of E-cadherin and ZO-1. Furthermore, these constructs promote albumin secretion, urea production, and P450 metabolic activity. In addition, the HCD hepatic tissue can inactivate YAP / TAZ pathway through cell-cell interactions, thereby maintaining primary hepatocyte functions. Further screening indicated that hepatocytes grown in the dense model were more sensitive to drug treatments compared to those in a lower density hepatic model, underscoring the importance of HCD in recapitulating physiological responses to drug compounds. Overall, the inventive approach represents a significant advancement in liver tissue engineering, providing a promising platform for the development of physiologically relevant in vitro liver models for drug screening and toxicity testing.

[0013] Biocompatible iodixanol (IDX) enables precise tuning of optical bioink properties, leading to minimal light scattering effects. Descriptions of the use of IDX for bioprinting can be found in International Publication No. WO 2024 / 118942, which is incorporated herein by reference. This approach achieves micron-scale resolution with a high density of cells encapsulated in natural hydrogels. Using the inventive approach, bioinks such as methacrylated gelatin (GelMA) and polyethylene glycol-norbornene (PEGNB) can be combined with a refractive index matching agent such as IDX to produce an HCD liver model using DLP bioprinting. Using the inventive approach, hepatic tissues composed of primary mouse hepatocytes (PMH) were bioprinted with high cell viability, facilitating accelerated tissue remodeling due to robust cell-cell interactions.

[0014] In an exemplary implementation, markers of cell-cell interaction, including albumin secretion, urea production, and CYP450, were highly expressed. By assessing transcriptome profiles, liver-specific markers, and metabolic gene expression, the HCD hepatic model was able to maintain hepatic function. Moreover, the HCD hepatic tissue creates a confluent cell environment that upregulates cell adhesive and tight junction gene expression, leading tothe inactivation of the YAP / TAZ pathway through cell-cell interactions, thereby preserving primary hepatocyte function. Finally, based on this HCD model, the metabolic transformation of various drugs was investigated via dynamic culture systems, utilizing CYP activity measurement, mass spectrometry analysis, and drug cytotoxicity assays to elucidate metabolic enzyme functions in hepatic tissue.

[0015] In another implementation, a bioprinted multi-cellular liver model with induced pluripotent stem cells (iPSCs)-derived hepatocytes and endothelial cells was integrated within a matrix metalloproteinase (MMP)-degradable, YIGSR-functionalized polyethylene glycol (PEG) — norbornene (NB) hydrogel. This multi-cellular architecture recapitulates hepatic organization by spatially positioning endothelial and parenchymal compartments in physiologically relevant arrangements, enabling paracrine signaling, enhanced nutrient exchange, and stable cell-cell / matrix interactions. Integrated with a dynamic microfluidic perfusion system, the construct supports matrix remodeling, delivers physiological shear stress, and sustains a well-oxygenated microenvironment. Compared to static culture, dynamic perfusion preserved long-term albumin and urea secretion, enhanced cytochrome P450 (CYP450) activity, reduced oxidative stress, and maintained mitochondrial integrity over extended culture periods. Transcriptomic profiling confirmed significant enrichment of metabolic, junctional, and drug-processing pathways. Functionally, the multi-cellular platform demonstrated robust and inducible drug-metabolizing capacity, enabling accurate identification of clinically relevant hepatotoxic compounds. Drug potency metrics — including ICso and benchmark dose values — closely matched reported human plasma concentration thresholds, underscoring the translational potential of the system.

[0016] The inventive scheme employs 3D bioprinting technology to fabricate high-cell-density liver acinus structures, incorporating iPSC-derived hepatocytes, liver sinusoidal endothelial cells (LSECs), hepatic stellate cells (HSCs), cholangiocytes, and immune cells. In some embodiments, these constructs are bioprinted in a 96-well plate format, allowing for parallel perfusion and real-time monitoring of drug metabolism and toxicity. The dynamic microfluidic system simulates blood flow and physiological shear stress, enhancing the functionality of the model. Al algorithms analyze datasets from high-content imaging, cytokine profiling, and molecular assays, generating actionable insights into drug performance and potential toxicity.

[0017] In one aspect, a multicellular liver model includes a construct 3D bioprinted from a bioink to form a high cellular density (HCD) liver acinus structure, wherein induced pluripotent stem cell (iPSC)-derived hepatocytes and supporting cells are incorporated into the construct during printing. In some embodiments, the supporting cells are one or a combination of liver sinusoidal endothelial cells (LSECs), hepatic stellate cells (HSCs), human umbilical vein endothelial cells (HUVECs), cholangiocytes, and immune cells. The construct may be bioprinted in a plurality of layers including a first layer corresponding to a hepatic chord structure and comprising iPSC-derived hepatocytes; and a second layer corresponding to a vascular layer and having a permeable interface configured for molecular exchange between a nutrient medium and the first layer. The construct may further comprise an inter-channel formed between the first and second layers, the inter-channel configured to create a metabolic gradient in a fluid flow through the construct. In some embodiments, the first layer is printed with a cell density within a range of 1 million cells / mL to 1,000 million cells / mL. In some embodiments, the first layer may be printed with a cell density within a range of 60 million cells / mL to 80 million cells / mL. The first layer may further include cholangiocytes. The second layer may include LSECs, HSCs, and / or HUVECs.

[0018] In some embodiments, the construct is bioprinted in one or more wells of a multiwell plate, which may be a 96 well plate. The bioink is a mixture of hydrogel and iodixanol (IDX). In some embodiments, the hydrogel is GelMA. In other embodiments, the hydrogel is polyethylene glycol (PEG) functionalized with norbornene groups norbornene (PEGNB). The PEGNB may be crosslinked using an MMP-degradable peptide as the cleavable crosslinker and functionalized with the CYIGSR peptide as a cell-adhesive ligand. A ratio of a non-degradable to a degradable crosslinker may be selected to control structural integrity of the construct over time. In some embodiments, the ratio is 40% degradable: 60% non-degradable.

[0019] In some embodiments, the iPSC-derived hepatocytes are patient-specific. The construct may be integrated with a microfluidic chamber, the microfluidic chamber configured for in-perfusion of oxygen and nutrients and out-perfusion of metabolic by-products through the liver acinus. The microfluidic chamber includes an inlet and an outlet, the inlet configured for introducing a sample to be tested.

[0020] In another aspect, a method for modeling liver disease and hepatotoxicity includes introducing a sample to be screened to the multicellular liver model; and monitoring the multicellular liver model for changes in expression of one or more genes indicative of reaction to the sample. The method may further include inducing different hepatic phenotypes using co-cultures of supporting cells.

[0021] The inventive approach provides four key innovations: (1) human iPSC-derived models enable the creation of patient-specific liver models, providing a personalized approach to drug screening; (2) 3D bioprinting reconstructs the liver acinus structure, replicating spatial multicellular interactions critical for liver physiology and drug metabolism; (3) high-throughput bioprinting techniques produce liver constructs in a 96-well chip format, with perfusable samples that allow real-time monitoring, significantly improving scalability and throughput for drug screening; and (4) Al-assisted data analysis integrates complex datasets to enhance the interpretation of drug testing results, providing deeper insights into drug behavior and toxicity.BRIEF DESCRIPTION OF THE DRAWINGS

[0022] FIG. 1A is a schematic illustration of a 3D bioprinting set-up for fabrication of an HCD liver model in a 96-well plate according to an embodiment of the inventive scheme;FIG. IB diagrammatically illustrates a bioprinted model in a 96-well plate and a schematic of perfusion within the model; FIG. 1C diagrammatically illustrates a multicellular liver acinus formed through precise positioning of multiple human iPSC-derived cell types in a structure that mimics the physiological cell-cell interactions in a native liver microenvironment.

[0023] FIGs.2A-2C compare characteristics of the inventive HCD liver model, where FIG.2A is a plot showing the relationship between scaffold compressive modulus and print exposure time for different cell densities; FIG. 2B is a quantitative plot showing the compressive moduli of cell-embedded scaffolds over 7 days; and FIG. 2C compares cell viability measurement of different hepatocyte densities at various time points.

[0024] FIGs. 3A-3F illustrate how bioprinted HCD hepatic model enhancing cell-cell interaction and maintains hepatic functions, where FIGs. 3A-3B plot flow cytometry analysis of intracellular albumin production; FIGs. 3C-3D show evaluation of the expression of % positive cells and quantitative analysis of E-cadherin and ZO-1, respectively, by flow cytometry; FIG. 3E plots albumin (left) and urea (right) secretion byhepatocytes across H, M, and L densities over time; and FIG. 3F compares CYP3A4 and CYP2C9 activity in different density groups at day 7, where error bars represent mean ± s.d. ***p < 0.001, **p < 0.01 and *p < 0.05.

[0025] FIGs.4A-4C illustrate transcriptomic analysis of PMH cultivated in HCD and MCD models, where FIG. 4A is a volcano plot of statistically significant differentially expressed genes at Q < 0.05 identified from RNA-Seq libraries of PMHs; FIG. 4B provides Top 10 GO terms of up-regulated (top) and down-regulated (bottom) DEGs; and FIG.4C compares PMH gene expression in high and medium culture densities. Error bars represent mean ± s.d. *p < 0.05 and **p < 0.01 .

[0026] FIGs. 5A-5D illustrate that recapitulating the physiological liver environment maintains hepatocyte enzyme functions and inactive YAP status in the HCD model, where FIG. 5A is a series of plots showing cytochrome P450 related CYP gene expression fold change of PMH for HCD and MCD; FIG. 5B is a schematic diagram of AJs protein E-cadherin and TJs protein ZO-1 by which mechanical cues inhibit the activities of YAP / TAZ;FIG. 5C provides plots showing AJs and TJs gene expression fold change of PMH; and FIG. 5D provides plots showing YAP related gene expression fold change of PMH. Error bars represent mean ± s.d. *P < 0.05 and **p < 0.01.

[0027] FIGs. 6A-6G illustrate results for the inventive HCD liver model for drug testing in a dynamic culture system, where FIG. 6A plots gene expression profiles showing levels of CYP3a41, CYP2c55, CYP2blO, and CYPla2 in untreated (CTL) and rifampicin (RIF) treated samples on day 7 across three density conditions, CTL (left bar), 20 pM RIF (center bar), and 40 pM RIF (right bar). Error bars represent SEM, and n = 3 was used for all data points; FIG. 6B shows CYP3A4 and CYP2C9 activity induction by RIF treatment across three density conditions (CTL (left), 20 pM RIF (center), and 40 pM RIF (right); FIG. 6C provides calibration curves of 4-OH-Diclofenac and 1-OH-Midazolam standards as dictated by mass spectrometry (fragment ion m / z 230.0 and m / z 203.0, respectively); FIG.6D shows the conversion ratio of diclofenac and midazolam metabolic products collected from cell culture medium after treating HCD and MCD constructs for 24 and 48 hours; FIG. 6E shows the cytotoxicity analysis of CPA in the HCD and MCD liver models. Error bars represent mean ± s.d. *p < 0.05, **p < 0.01 and ***p < 0.001; FIG. 6F plots cytotoxicity analyses of CPA in HCD and MCD liver models in a static culture; and FIG. 6G plots CYP3 A4 upregulation in HCD and MCD liver models after CPA treatment.

[0028] FIG. 7 diagrammatically illustrates formulation of a PEGNB-based hydrogel according to an embodiment of the inventive liver model in which PEGNB is crosslinked using an MMP-degradable peptide as the cleavable crosslinker and functionalized with the CYIGSR peptide as a cell-adhesive ligand.

[0029] FIGs. 8A-8E illustrate formulation, optimization, and characterization of a celladhesive, photoconjugatable, and degradable PEGNB-based hydrogel for liver tissue engineering, where FIG. 8A plots viability of iPSC-Heps encapsulated in PEGNB hydrogels functionalized with different concentrations of CYIGSR peptide, evaluating the hydrogel’s cell-adhesive capacity; FIG. 8B show fluorescence intensity of FAM-labeled YIGSR peptide (FAM-YIGSR) within the PEGNB hydrogel, assessing photoconjugation efficiency and uniformity of peptide distribution following crosslinking; FIG. 8C plots stability and enzymatic degradability of D hydrogels in phosphate-buffered saline (PBS) and collagenase solution (4 mg / mL) under physiological conditions; FIG. 8D shows quantification of pore size in degradable (D) and non-degradable (ND) PEGNB hydrogels following collagenase treatment over time; and FIG. 8E plots time-dependent stiffness measurements of iPSC-hepatocyte-laden D and ND hydrogels over a 3-week culture period, evaluating hydrogel remodeling and maintenance of mechanical integrity.

[0030] FIGs. 9A-9E provide a comparative analyses of hepatocyte network formation, function, and transcriptomic profiles in degradable (D) versus non-degradable (ND) PEGNB hydrogels, where FIG.9A is a schematic representation of iPSC-heps organization D) and ND PEGNB hydrogels; FIG. 9B shows results of quantitative analysis of E-cadherin- and ZO-1 -positive cells, expressed as the percentage of marker-positive cells per field; FIG. 9C shows Cytochrome P450 activity assays for CYP3A4 and CYP2C9 in hepatocytes culture, comparing D and ND conditions; FIG. 9D provides results of functional assessment of hepatic metabolism by quantifying albumin and urea secretion in D and ND hydrogels; and FIG. 9E provides comparative gene expression profiles for key markers associated with cell-cell / ECM interactions and hepatic function in D and ND hydrogels.

[0031] FIGs. 10A-10D show functional and transcriptomic comparison of a multilayer liver co-culture model and a monolayer hepatocyte culture, where FIG. 10A plots results of functional assessment of hepatic metabolism by quantifying albumin and urea secretion in CO and MONO cultures, FIG. 10B plots cytochrome P450 activity assays for CYP3A4,CYP2B6, CYP1 A2, and CYP2C9 in CO (left bar) and MONO (right bar) culture; FIG. IOC provides Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) in the biological process (BP) and molecular function (MF) categories; FIG. 10D provides fold-change plots of cytochrome P450 (CYP) family genes and phase II enzyme genes in CO relative to MONO liver models.

[0032] FIGs. 11A-11F Provide comparative functional, metabolic, oxidative stress, and transcriptomic characterization of dynamic perfusion versus static culture conditions in a 3D liver model, where FIG. 11A provides results of functional assessment of hepatic metabolism by quantifying albumin and urea secretion in dynamic perfusion and static culture conditions; FIG. 11B plots cytochrome P450 activity assays for CYP3 A4, CYP2B6, CYP1A2, and CYP2C9 under dynamic and static conditions; FIG. 11C plots measurement of reactive oxygen species (ROS) generation, expressed as DCF fluorescence intensity, in dynamic and static liver models over time, demonstrating reduced ROS accumulation in dynamic culture; FIG. 11D-11E provide fold-change plots of genes associated with oxidative stress response and, inflammatory processes, respectively; and FIG. 11F plots relative expression of cytochrome P450 (CYP) family enzymes in dynamic and static liver models over time.

[0033] FIGs. 12A-12E provide functional characterization and non-hepatotoxic drug toxicity evaluation of the matured bioprinted coculture liver model cultured under static or perfused condition, where FIG. 12A plots results of quantification of metabolic conversion efficiency after 48 hours of culture in static or dynamic conditions; FIG. 12B and 12C provide relative mRNA expression levels and activity, respectively, of cytochrome P450 (CYP) isoforms and corresponding enzyme activity following drug-mediated induction under static and dynamic conditions; FIGs. 12D and 12E provide results for drug-induced cytotoxicity analysis in the bioprinted co-culture liver model following exposure to non-hepatotoxic reference compounds: aspirin and dexamethasone, respectively, for 24 and 48 hours under static and dynamic conditions followed by estimation of cell viability, DNA content, and intracellular LDH activity. All data are presented as mean ± S.D. (n = 5); where statistically significant difference is represented by *p < 0.05, **p < 0.01 and ***p < 0.001.

[0034] FIGs. 13A-13E provide drug toxicity prediction in the matured bioprinted liver model exposed to hepatotoxic drugs, acetaminophen (APAP) (FIG. 13A), valproate (VLP) (FIG. 13B), voriconazole (VRZ) (FIG. 13C) and troglitazone (TGZ) (FIG. 13D), in staticor dynamic condition for 48 hour followed by estimation of cell viability, DNA content and LDH activity; and FIG. 13E plots regression analysis of the relationship between IC50 or BMD values with Cmax.DETAILED DESCRIPTION OF EMBODIMENTS

[0035] According to embodiments disclosed herein, hydrogel scaffolds are fabricated to produce an HCD liver model using DLP bioprinting. These HCD liver acinus structures, corresponding to the smallest functional unit of the liver, include hepatic parenchymal cells, primarily hepatocytes, and non-parenchymal cells such as liver sinusoidal endothelial cells, organized within a specialized vascular network. The structures were bioprinted in different hydrogel scaffolds, first using primary mouse hepatocytes (PMH) and then using co-cultures of human iPSC-Heps and HUVECs (Human Umbilical Vein Endothelial Cells), with high cell viability, facilitating accelerated tissue remodeling due to robust cell-cell interactions.

[0036] Referring to FIG. 1A, three-dimensional (3D) hydrogel scaffolds were fabricated using a 3D bioprinter under ambient room temperature conditions. Descriptions of the basic bioprinter elements have been previously provided. See, for example, International Publication No. WO 2024 / 118942. Briefly, the bioprinter 100 includes a UV source 102, a digital micromirror device (DMD) 106 which modulates light impinging on the DMD using a series of masks 104 to project patterns to construct layers of the scaffold. System optics 108 focus the modulated light to an x, y, z stage on which a photopolymerizable hydrogel is placed for exposure. Referring to FIG. IB, individually perfusable chambers may be printed within each well 112 of the multi-well plate 110. For evaluation, a 96-well format chip (85.48 x 127.76 mm) was used to construct a microfluidic chip, where each well will contain an individually perfusable liver acinus model, each with at least one inlet 122 and outlet 124. Image 120 corresponds to fluorescence and bright-field images (5x) of the acinus model showing the spatial distribution of fluorescently labeled iPSC-Hep (gray areas) and HUVECs (dark areas) within degradable PEGNB hydrogels in the co-culture (CO) model. Scale bar: 100 pm.

[0037] The body of the chip is manufactured using PDMS, with the bottom assembled to a No. 1.5 coverslip (0.17 mm thickness) for high-resolution imaging. The PDMS surface may be treated with lipophilic coatings to avoid small molecule absorption. Each microfluidic chamber (6x2x0.5 mm) is primed hydrophilically for perfusion, with each chamber connected to perfuse-in and perfuse-out systems, driven by a microfluidic pump 126.Microfluidic devices are well known in the art and widely commercially-available. Accordingly, details of the interface architecture, pump features, and other components of a microfluidic system will not be described herein.

[0038] A high-speed, high-precision 3D bioprinter (BIONOVA X, Cellink), was used to print the liver model directly in the 96-well chip. Scaffold geometries were digitally designed using computer-aided design (CAD) software with the BIONOVA X slicing platform. The slicing parameters, including layer height, print path, and infill density, were adjusted based on the desired scaffold architecture. Printing was conducted directly onto methacrylated glass coverslips in the well plate, which were functionalized to enable stable anchoring of the photocrosslinked hydrogel constructs. Photocrosslinking was achieved using a built-in 405 nm blue LED light source at an intensity of 12 mW / cm2, corresponding to 75% of the printer’s maximum output. This controlled exposure facilitated rapid and uniform crosslinking of each printed layer. For multi-cellular printing, each printed zone had an identical thickness. After printing the first layer, excess uncrosslinked hydrogel precursor was gently removed by rinsing with pre-warmed Dulbecco's phosphate-buffered saline (DPBS, 37 °C) to prevent mixing or deformation before printing the subsequent layer. The second zone was then deposited with the same thickness as the first, ensuring consistent structural dimensions throughout the scaffold.

[0039] Markers of cell-cell interaction, including albumin secretion, urea production, and CYP450, were highly expressed. By assessing transcriptome profiles, liver-specific markers, and metabolic gene expression, the inventive HCD hepatic model was confirmed to maintain hepatic function. Moreover, the HCD hepatic tissue creates a confluent cell environment that upregulates cell adhesive and tight junction gene expression, leading to the inactivation of the YAP / TAZ pathway through cell-cell interactions, thereby preserving primary hepatocyte function. Finally, based on this HCD model, the metabolic transformation of various drugs was investigated via dynamic culture systems, utilizing CYP activity measurement, mass spectrometry analysis, and drug cytotoxicity assays to elucidate metabolic enzyme functions in hepatic tissue.

[0040] To achieve the desired liver model, bioinks should preferably be optimized for each liver cell type. A key element of successful bioprinted liver constructs lies in the design of hydrogel bioinks, which must be mechanically tunable, cytocompatible, and permissive to cell-mediated remodeling. While natural hydrogels such as collagen, gelatin, and alginateoffer inherent bioactivity, they may suffer from batch-to-batch variability and limited mechanical tunability, which compromise reproducibility and scalability. On the other hand, synthetic hydrogels — particularly those based on polyethylene glycol (PEG) — offer a highly modular and reproducible platform. Among these, 8-arm PEG functionalized with norbomene groups (PEGNB) is particularly well-suited for DLP printing due to its compatibility with thiol-ene photopolymerization, enabling rapid and uniform crosslinking under cell-friendly conditions.

[0041] A liver acinus structure is bioprinted by employing a two-step printing strategy to closely replicate the intricate homotypic and heterotypic cellular interplays in vivo. The hepatic cord design is not merely structural; it is intentionally engineered to recreate the liver acinus, the smallest functional unit of the organ. An inter-channel space between the vascular and parenchymal layers is designed to establish a metabolic gradient, enabling zonation-like behavior. This as biomimetic zonation is controllable, so that the platform does not simply support cell growth, but actively replicates the spatial multicellular interactions and fluid dynamics characteristic of a native liver sinusoid. Referring to FIG. 1C, the hepatic cord structure is bioprinted as a first layer (0.15x1x0.25 mm), consisting of iPSC-derived hepatocytes and cholangiocytes. Then the ‘Space of Disse’ layer containing iPSC-derived LSECs, and HSCs is printed as a second layer (0.02x1x0.25 mm). The hepatocyte regions may be printed with a cell density within a range of 1 to 1,000 million / mL. The vascular layer is printed at a slightly lower cell density to allow for optimal post-printing angiogenesis. The interval (e.g., d = 40 pm) between the endothelial layers should mimic the lumen space in the liver sinusoid. Within the bioprinted liver acinus, hepatocytes recreate the liver cords, performing essential metabolic and detoxification functions, similar to their role within the liver sinusoids in vivo. LSECs form a permeable interface, enabling selective molecular exchange between the medium and hepatocytes, while HSCs deposit ECM and mediate responses to liver injury.

[0042] Once bioprinted, the liver acinus can be integrated with a microfluidic chip and perfused using an automated micro-pump. The perfusion system passes the medium through the 40 pm inter-channel between the endothelial layers, creating a metabolic gradient from the upstream portal zone to the downstream central zone. The flow rate is controlled to replicate the blood flow through the liver sinusoid. Finally, immune cells, including iPSC-derived KCs, B cells, T cells, and NK cells, can be incorporated into the inter-space of theliver sinusoid through medium perfusion. This allows recreation of the immune microenvironment of the liver tissue, capturing both the metabolic and immune dynamics of the native liver. For creation of a patient specific liver model, autologous iPSCs may be used.

[0043] Physiological hemodynamics, including shear stress and fluidic pressure, are essential for maintaining the functional integrity of the liver microenvironment. Physiological flow rate ensures that applied risk factors, such as toxins or pathogens, penetrate the endothelial layer and interact with the liver parenchymal cells similar to in vivo conditions. To achieve this, the + speed is optimized to mimic the hemodynamics in human liver sinusoid, such as the shear stress below 10 dyne / cm2, and the flow rate ranging from 2.59 to 400 pm / s. Computational Fluid Dynamics (CFD) may be used to simulate the perfusion in the bioprinted liver system and predict the shear stress and pressure distribution throughout the liver acinus. Factors such as channel geometry, fluid viscosity, and flow rates are considered in the simulation to optimize the perfusion settings. The goal is to maintain physiological levels of shear stress that support the cellular function without causing disruption or detachment of immune cells. Physics-guided deep learning can be applied to analyze experimental data and refine the perfusion conditions, ensuring the dynamic behavior of the immune cells and the proper delivery of risk factors.

[0044] As described in the following sections, different bioinks were evaluated for use in bioprinting of the liver model. A first implementation, described in Example 1 below, focused on bioinks based on gelatin methacryloyl (GelMA) and collagen, which have demonstrated excellent printability in multicellular iPSC-derived liver models and have maintained hepatocyte function for up to two weeks post-bioprinting. To enhance vascularization and angiogenesis, the optimized bioink may include ECM components such as collagen type IV, laminin, and fibronectin. Evaluation standards will assess bioink suitability for 3D liver tissue bioprinting, including printability, mechanical properties, and cell viability after printing.Example 1: GelMA-Based Bioinks

[0045] GelMA synthesis: Gelatin methacrylate (GelMA) was synthesized from type A gelatin (Cat. #G2500, Sigma-Aldrich) and methacrylic anhydride (MA, Cat. #276685, Sigma Aldrich) in a 0.25 M carbonate-bicarbonate (CB) buffer, following a previously described method. In brief, a 10% (w / v) gelatin solution in the CB buffer was allowed toreact with MA at a MA / gelatin ratio of 0.085 mL / g for 1 h at 50 °C. The reaction was then quenched by adjusting the pH to 7.4. The resulting mixture was dialyzed using 12-14 kDa cutoff dialysis tubing against MilliQ water for 3 days and subsequently lyophilized for future use. The degree of functionalization (DoF) of GelMA was quantified by 1H-NMR.

[0046] Primary mouse hepatocyte isolation: All research conducted in this study adhered to approved Institutional Review Board (IRB) protocols at the University of California, San Diego (UCSD). Furthermore, all animal work associated with this research received approval from UCSD’s IRB. The animal work was conducted in strict accordance with the guidelines outlined by the Institutional Animal Care and Use Committee (IACUC), ensuring ethical and humane treatment of animals. Primary mouse hepatocytes (PMH) were isolated from 8- to 10-week-old mice using a two-step collagenase digestion method. In brief, mice were euthanized with CO2, and the abdomen was carefully opened to expose the portal vein and inferior vena cava. Perfusion buffer was introduced into the portal vein, and the vena cava was severed. The liver was perfused with the buffer, which contained collagenase H (Cat. #11074032001, Roche) and HEPES (Cat. #15630080, Gibco), at a rate of approximately 10 mL / min. The liver was then carefully excised and washed through a 100 pm filter. Hepatocytes were isolated by gradient centrifugation. The resulting single-cell suspensions were washed with DPBS, and the cells were resuspended in hepatocyte culture medium (HCM; Cat #CC-3198, Lonza) containing all supplied supplements along with 1% penicillin-streptomycin.

[0047] 3D Bioprinting of the PMH models: GelMA and lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP, Cat. #85073-19-4, Sigma-Aldrich) were dissolved in Dulbecco's phosphate-buffered saline (DPBS, Cat. #14190144, Gibco) to create stock solutions of 20% (w / v) and 4% (w / v), respectively. To prepare the prepolymer solutions, all stock solutions were warmed to 37 °C prior to use. The final prepolymer solution was formulated by mixing 12% GelMA, 0.4% LAP, 60% iodixanol (IDX), and DPBS as the solvent. This solution was maintained at 37°C on a heat block and was covered with aluminum foil to prevent premature photopolymerization. Micro-tissue constructs were bioprinted using a previously reported method. In brief, an in-house micro-continuous optical printing (pCOP) system was employed, and polydimethylsiloxane (PDMS) spacers were utilized to control sample thickness. A PDMS-coated coverslip was placed on top of the solution and spacers to create a smooth surface. The solution was then exposed to UVlight using a projected light pattern. For bioink preparation, the prepolymer solution was combined with the cell suspension at a 1 : 1 ratio, resulting in final cell densities denoted high (80 million cells / mL), medium (20 million cells / mL), or low (5 million cells / mL), depending on the printing requirements. The cell-laden bioink was then carefully loaded onto the printer stage for use. After loading, bioink was polymerized over an appropriate exposure time using 365 nm light. Following printing, warmed DPBS was used to wash away the excess bioink-cell solution.

[0048] Mechanical testing: A mechanical tester (“MicroSquisher,” CellScale) was employed to measure the compressive modulus of the printed samples. For each cell-bioink mixture, pillars with a diameter and height of 500 pm were printed using the same setup used for different cell density liver models. The pillars were stored at 37 °C prior to taking measurements to mimic physiological culture conditions. While taking measurements with the MicroSquisher, stainless steel beams and platens were used to consecutively compress the constructs by 10% of their original height for three cycles. Data from the last of the three measurements was used for analysis, and the compressive modulus was calculated using customized MATLAB scripts to analyze the force and displacement data.

[0049] Cell viability analysis: Cell viability analysis was conducted on days 1, 3, 7, and 10 following bioprinting with PMHs. To conduct live / dead staining, samples were washed twice with DPBS after removing the culture medium. Samples were subsequently incubated with a solution containing 1 pM calcein AM (live cell stain, Cat. #C3099, Invitrogen) and 2 pM ethidium homodimer- 1 (dead cell stain, Cat. #P3566, Invitrogen) at room temperature for 30 min. Image acquisition using a Leica DMI 6000B microscope (Leica Microsystems) was immediately performed after incubation. Cell viability was also quantified using the CCK-8 assay (Cat. #K1018, ApexBio) using four biological replicates for each condition. At designated time points, the samples were washed with DPBS, incubated with 500 pL of fresh media with 10% CCK-8 reagent under regular incubation conditions for 60 min. After incubation, 200 pL of the supernatant was collected from each sample, and its absorbance was measured at 450 nm using a Tecan Infinite 200 PRO.

[0050] Immunofluorescence staining: Bioprinted constructs were rinsed with Dulbecco's phosphate-buff ered saline (DPBS) three times and fixed with 4% paraformaldehyde for 1 hour at room temperature. The block / permeabilization solution was prepared by dissolving 5% (w / v) bovine serum albumin (BSA, SKU#700-101P, GeminiBio) and 0.1% Triton X-100 (Cat. #H5141, Promega) in DPBS. The solution was filtered after complete dissolution. Fixed samples were blocked / permeabilized for 1 hour at room temperature on a shaker at 100 rpm. Primary antibodies, listed in Table 1, were diluted in Cell Staining Buffer (Biolegend), and the samples were incubated in the primary antibody solution overnight at 4 °C.Table 1Primary antibodies IF staining Flow Cyt [EPR20195]# “tlbOdyAbeam ab207327 1:500 1:200E-Cadherin (Alexa Fluor® Cell nomn >488 C.onj •uga ;te x) signa rling #86770 1:50 1:50ZO-1 (D6L1E) Rabbit mAb(Alexa Fluor® 647 Abeam ab221547 1:100 1:50 Conjugate)Secondary antibodiesGoat Anti-Rabbit (Alexa Abeam abl50084 1:200 1:1000Fluor® 594)

[0051] Subsequently, samples were rinsed three times with DPBS and incubated with secondary antibodies and DAPI in the dark for 1 hour at room temperature on a shaker. After the secondary incubation, samples were rinsed three times with DPBS. The stained samples were then imaged using a Leica SP8 fluorescence confocal microscope.

[0052] Flow cytometry: To analyze different PMH densities in the bioprinted construct, cells were collected by degrading the collagen matrix with collagenase type I (2 mg / mL, Cat. #17100-017, GIBCO) at 37 °C for 30 min. The collected cells were washed, fixed, and blocked using the Cytofix / Cytoperm kit (Cat. 554724, BD Biosciences), followed by additional blocking with TruStain FcX PLUS (Cat. #156603, Biolegend,). Primary antibodies (refer to Table 1) were used for cell staining at 4 °C for 30 min. After three washes, cells were stained with secondary antibodies. Flow cytometry was performed using a BD LSRFortessa SORP. All data were analyzed using FlowJo vlO.

[0053] Cytochrome P450 activity: Cytochrome P450 activity was assessed using a P450-Glo 2C9 / 3A4 Assay Kit (Promega), and the luminescence activity was measured with a luminometer in accordance with the manufacturer’s instructions. 3D printed liver models were washed in DPBS before being incubated with reagent for 1 hour at 37 °C. After washing, 25 pL of supernatant was transferred into the wells of a white 96-well plate, mixedwith 25 pL of the detection reagent, and incubated for 20 min in the dark at room temperature. The plate was transferred into a multimode microplate reader and luminescence was measured using a 1 s integration time. The background signal of the assay (no cells) was subtracted from measurements of the cell cultures.

[0054] Albumin and urea secretion quantification: The culture supernatants were collected from samples at various time points and stored at -80 °C before analysis. Urea production quantification was performed using a QuantiChrom™ BCG Albumin Assay Kit and QuantiChrom™ urea assay kit (BioAssay Systems) according to the manufacturer’s instructions. The amounts of albumin and urea secreted were calculated based on standard curves from each experiment, followed by normalization to the viable cell population, ensuring accurate representation of secretion data relative to the actual number of active cells within the constructs.

[0055] RNA isolation and quantitative real-time reverse-transcription PCR (qRT-PCR):RNA extraction was conducted using TRIzol reagent (Life Technologies). To obtain sufficient RNA from the bioprinted liver constructs, 10 samples were combined and all RNA was extracted in unison. In brief, bioprinted liver constructs were incubated with TRIzol reagent followed by purification using a spin-column method with Direct-zol RNA Microprep (Cat. #R2060, Zymo Research), and the RNA concentration was evaluated using a Tecan plate reader after resuspending RNAs in RNase-free water. The RNA samples were immediately stored at -80 °C. For qRT-PCR, 200 ng of RNA was first converted to cDNA using the First Strand cDNA Synthesis Kit (Cat. #E6300S, New England BioLabs). Subsequently, qRT-PCR was performed using the Luna® Universal qPCR Master Mix (Cat. #M3003S, New England BioLabs) and the Quantstudio 3 qRT-PCR system. The relative quantification of specific genes was determined by normalizing the threshold cycle (Ct) values against the housekeeping gene. The forward and reverse primers were purchased from Integrated DNA Technologies, and primer sequences are listed below in Table 2.Table 2Primer Primer sequence (5’-3’)Alb F GGCTACAGCGGAGCAACTGAR GCCTGAGAAGGTTGTGGTTGTGHnf4a F TGCGAACTCCTTCTGGATGACCR CAGCACGTCCTTAAACACCATGGSerpinfl F TGCTTCTCCTCAACGCCATCCAR ATGTCCACCGACACTGTGAACC Tat F GTCGCTTCCTATTACCACTGTCC R CGGAATGAGGATGTTCTGTCCAGAhsg F CACACTGGAGACCACTTGCCAT R GCCGTCTTGTTTCAGGATGTGGTtr F GAGTAGAACTGGACACCAAATCG R CTGCGATGGTGTAGTGGCGATGAsgr2 F GCTCTTTCACCTGAAGCACTTCC R CCCGAGAAAACCAGTAGCAGCTAfp F GCTCACATCCACGAGGAGTGTT R CAGAAGCCTAGTTGGATCATGGG CYPla2 F ATAACTTCGTGCTGTTTCTGC R ACCGCCATTGTCTTTGTAGT CYP7al F CACCATTCCTGCAACCTTCTGG R ATGGCATTCCCTCCAGAGCTGA CYP2el F AGGCTGTCAAGGAGGTGCTACT R AAAACCTCCGCACGTCCTTCCACyp3a41a F TGCTCTTCACCATGACCCACAG R CCTCATGCCAATGCAGTTCCTGCyp2c55 F AGGAAGCTCTGGATGACCTTGG R TTGAGAAGCGCCGAAGCTCCTTCyp2bl0 F TTCTGCGCATGGAGAAGGAGAAGT R TGAGCATGAGCAGGAAGCCATAGTCdhl F GGTCATCAGTGTGCTCACCTCT R GCTGTTGTGCTCAAGCCTTCACTjpl F GTTGGTACGGTGCCCTGAAAGA R GCTGACAGGTAGGACAGACGATYap F GCCATGTTGTTGTCTGATCGR TGCCATGTGGTGATTTTCTCWwtrl F TGCCATGTGGTGATTTTCTCR CCTATGACGTGACCGACGAGCyr61 F TTTACAGTTGGGCTGGAAGC R CACCGCTCTGAAAGGGATCTCtgf F GCTTGGCGATTTTAGGTGTC R CAGACTGGAGAAGCAGAGCC GAPDH F CATCACTGCCACCCAGAAGACTG R ATGCCAGTGAGCTTCCCGTTCAG

[0056] RNA sequencing and data analysis: RNA-seq was performed on the following samples: (1) HCD and (2) medium cell density (MCD) groups. Total RNA was isolated by the extraction reagent TRIzol introduced above. The concentration of the RNA library was diluted to 1 ng / pL according to the measurement by Qubit® RNA Assay Kit in Qubit® 3.0 and the insert size was assessed by the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). Cluster generation and sequencing were conducted with a Novaseq 6000 S4 platform. Around 40 million to 60 million total reads were obtained for each sample. RNA quality evaluation and sequencing were performed by Novogen Inc. For three biological replicates of each condition. For the analysis of Differentially Expressed Genes (DEGs), HISAT2 v2.0.5 was used to map sequence reads to the Mus musculus transcriptome (Mus musculus. mm39). Gene expression was evaluated by Fragments Per Kilobase Million Mapped Reads (FPKM) values by using HTseq-count (v.0.6.0). DESeq2 (v.1.20.0) was used to identify DEGs with a fold change cut-off threshold > 1 and false discovery rate (FDR)-adjusted p-value < 0.05. Typical DEGs related to the specific hepatic function and liver fibrosis were used for plotting heatmaps. Fold change of gene expression was normalized through logarithmic transformation at a base number of 2 as log2-FoldChange. Heatmaps of log2-FoldChange were plotted via SRplot heatmap.

[0057] Liver constructs within a microfluidic chip for dynamic culture: The uncoated p-Slide III 3D Perfusion chip (Ibidi, Cat. No: 80376) was selected for this study. Microfluidic experiments were performed using hepatocyte culture medium as previous described. Chips were pre-warmed for about 1 hour before loading. The 5 mm coverslip with a printed liver model was placed on the chip and 30 mL of culture medium was added. Lastly, the chip was sealed with an adhesive Ibidi Polymer Coverslip to prevent bubble formation. The dynamic culture condition was performed at a flow rate of 0.1 mL / min and a temperature of 37 °C.

[0058] CYP induction: To assess CYP induction potential, the CYP activity levels in rifampicin-treated and untreated samples were measured using the qRT-PCR and P450-Glo CYP Assay Kit as described above. The cells were treated with rifampicin (20 and 40 pM with a total of 0.1% DMSO in culture medium; Sigma- Aldrich) for 2 days, while non-treated samples (with a total of 0.1% DMSO in culture medium; Life Technologies) were used as controls for comparison. Relative quantification was performed using a standard curve, and the values were normalized against the input determined for the housekeeping gene GAPDH.

[0059] Mass spectrometry (MS) analysis for quantification of cellular drug metabolism:Phenotype-specific drug response of individual groups was assessed by subjecting cultures to a mixture of known substrates for the Cytochrome P450 system (Table 3). The collected medium was vortexed with methanol in a 1:2 volume ratio, and then the mixture was centrifuged at 10,000 rpm for 10 min to facilitate desalination and deproteinization. Metabolite concentrations were then measured via high performance liquid chromatography separation paired with tandem mass spectrometry detection on a Micromass Quattro Ultima triple quadrupole mass spectrometer. The supernatant was diluted with ultrapure water in a 1 : 8 volume ratio and loaded onto a C 18 column with an inj ection volume of 2 pL. Separation of drugs and their metabolites was performed on Cl 8 column at a flow rate of 3 pL / min and a temperature of 40 °C. Compound detection was based on the mass numbers detailed in Table 3 below, which provides the parameters for LC-MS / MS conversion rate quantification of drug metabolites. Based on the quantified metabolite concentrations, metabolic conversion rates were then estimated.Table 3Target Substrate Metabolite concentration m / zStandard ImipramineCYP2C9 Diclofenac 4-OH-Diclofenac 1 pM 312.0 ^ 230.0 CYP3A4 Midazolam 1-OH-Midazolam 2 pM 342.0 ^ 203.0

[0060] Drug response assessment: To evaluate the performance of the HCD liver model as a drug screening platform, a cytotoxicity assay was conducted on the chip. For comparison, an MCD model was also used. Both models were supplemented with increasing concentrations of Cyclophosphamide (CPA, Cat. #6055-19-2, Sigma) (5, 10, 50, 100, and 500 pg / mL). Viability was assessed after 24 hours using a CCK-8 test. Absorbance results at a wavelength of 450 nm were recorded for each concentration in each model.

[0061] Statistical analysis: Sample populations were compared using a t-test or one-way ANOVA performed with GraphPad Prism (GraphPad Software). *p < 0.05, **p < 0.01 and ***p < 0.001 were used as the thresholds for statistical significance. Data points on the graphs represent mean values with error bars representing SEM.Bioprinted liver models of different cell densities

[0062] To understand the effect of cell density on PMH function, three cell densities: 5, 20, and 80 million cells / mL, were selected to represent low (L), middle (M), and high (H) density groups, respectively.

[0063] Although low-density levels are commonly used to create functional constructs in many studies, it was hypothesized that such densities may be insufficient for generating highly functional tissue. An HCD bioprinting protocol was previously defined by incorporating a refractive index matching agent into the bioink, allowing the H group to be printed in the desired structure with minimal light scattering (International Publication No. WO 2024 / 118942). Varied IDX concentrations alter the refractive index of the bioink, allowing it to closely match that of the cytoplasm (1.36-1.39). A linear increase in refractive index was observed with increasing IDX concentrations up to 35%. This fine control is useful in minimizing light scattering, especially in high cell density larger constructs. While previous studies have employed IDX for scattering correction, a more precise refractive index tuning method enables critical enhancements in printability and optical transparency for engineered hepatocyte models. While numerous factors impact cell viability, results showed a mere 3.85% viability decrease after printing, attesting to the method’s reliability. Before investigating the effect of cell density on hepatic function, the printing parameters were adjusted to achieve consistent stiffness across groups with different cell densities. By varying the exposure time regionally, mechanical properties could be uniformly altered within the same construct. Mechanical testing of the 6% (wt / vol) GelMA-based constructs demonstrated a positive linear relationship between the stiffness and the exposure time. Emerging evidence suggests that mechanical stiffness, when mimicking native tissue environments, can impact hepatic function. Therefore, a photopolymerized matrix stiffness was selected similar to that of healthy liver tissues to support PMH (FIG.2A). Specifically, exposure times of 15, 20, and 40 seconds were chosen to produce scaffolds with identical stiffness (4 kPa) for the L, M, and H cell density groups. Cell-embedded scaffolds were fabricated under these conditions, and stiffness measurements were performed to determine scaffold stability over a 7-day culture period. The changes in stiffness over 7 days were not significant for all three conditions in cell-embedded scaffolds (FIG.2B). Post-printing, cell viability and live / dead staining showed that HCD bioprinting of PMHs achieved high viability (-80%) over the 7-day culture period. This indicates that HCD bioprinting cansustain cell viability over an extended duration, compared to hepatocytes cultured in a LCD (5 million cells / mL) format (FIG. 2C).Measuring metabolic function in the HCD hepatic model

[0064] To better understand cellular activity after bioprinting, intercellular interactions and metabolic hepatocyte functions were characterized. The underlying premise is that hepatocytes in HCD environments more readily form cell aggregates, which have been shown to enhance cell-cell interactions, cell viability, and metabolic function for prolonged periods due to improved retention of in vivo hepatic morphology. First, the size of cellular aggregates were measured, with the observation that larger aggregates in the HCD group formed after 7 days of culture. In contrast, only a few small aggregates formed in the medium cell density (MCD) group, and scarcely any formed in the LCD group. To study PMH in different cell density groups, immunofluore scent staining was performed for (1) E-cadherin (adhesive junction), an epithelial marker that had been shown to protect primary hepatocytes from apoptosis; (2) ZO-1 (tight junction); and (3) albumin, an indicator of hepatic function. HCD cultures after 7 days showed strong positive staining for all three markers, indicating functional spheroid formation. Quantifying fluorescent intensity in immunofluorescent images showed significant improvements of cell-cell interaction and albumin secretion in HCD compared to MCD and LCD. To further investigate cellular function, flow cytometry was used to measure intracellular albumin expression, which was found to be significantly higher in HCD cultures (FIGs.3A-3B). Notable changes were also observed in the expression of E-cadherin and ZO-1 between the different cell density groups (FIG. 3C-3D) Next, anabolic and catabolic function of the PMHs across different models were compared. Albumin secretion levels were monitored over a 2-week period, given its role in maintaining oncotic pressure in the body (FIG. 3E). Although albumin secretion declined in all three models after their respective peaks, PMHs in HCD maintained the highest level of albumin secretion 7 days after bioprinting. Similarly, urea production was measured over time, as it is indicative of the hepatocytes’ ability to break down amino acids. PMHs in HCD consistently exhibited higher urea production compared to those in MCD and LCD environments. The metabolic activity of hepatocytes, which largely depends on cytochrome P450 enzyme activity, was also assessed. Specifically, the activities of CYP2C9 and CYP3 A4 enzymes were measured — both were found to significantly increase in HCD cultures (FIG. 3F).Enhancement of liver-specific gene expression and functions of PMH in the HCD model

[0065] The aforementioned results discussing cell viability, cell-cell interaction, and liver metabolic function show that these cellular benchmarks dropped significantly after 7 days of culture in LCD conditions. This demonstrates that using current cell density ranges (1-10 million cells / mL), it is extremely difficult to maintain proper primary hepatocyte function. To further investigate the differences in cellular function under different cell density conditions, global transcriptome profiling was performed. RNA was extracted from PMHs cultured in bioprinted constructs and isolated from both HCD and MCD conditions. RNA sequencing (RNAseq) was then conducted to evaluate gene expression differences and identify major signaling pathways that were differentially regulated between the two conditions.

[0066] To assess the overall similarity between the transcriptional profiles of hepatocytes in HCD and MCD conditions Spearman rank correlation was used, yielding a correlation coefficient of 0.77, indicating a moderate degree of difference between the two conditions. Principal component analysis (PCA) was conducted to visualize the variance in gene expression between HCD and MCD. The PCA plot showed that the HCD culture had a markedly different transcriptional profile compared to the MCD condition, highlighting the significant impact of cell density on gene expression. To identify specific genes that were differentially expressed between the HCD and MCD conditions, a volcano plot shown in FIG. 4A was generated. This plot revealed a total of 7363 differentially expressed genes (DEGs) with a | log2(fold change) | and Q value < 0.05. These DEGs were labeled on the map, illustrating the extensive transcriptional reprogramming induced by the different cell density environments.

[0067] For analysis of liver-specific gene expression (LiGEP), homologous mouse LiGEP heat maps revealed an upregulation in gene expression profiles under HCD conditions on day 7, indicating good maintenance of liver function in HCD cultures. These results demonstrated effective preservation of PMH functionality under the described HCD bioprinting conditions. As represented by top 10 GO terms of DEGs (H vs M) in FIG. 4B, the upregulated GO terms were primarily associated with metabolic and biosynthetic processes, including cytochrome P450, steroid, and retinol metabolism, as well as cell-cell communications and interactions, such as cell junction organization and integrin signaling. These findings highlight the enhanced metabolic activity and robust cell-cell interactions inHCD cultures. Down-regulated DEGs largely pertained to cell division and cell cycle processes. While this suggests reduced cell proliferation under HCD conditions, this reduction also indicates that hepatocytes in HCD maintain normal physiological functions without initiating cell division and repair programs, which is consistent with a stable hepatic phenotype.

[0068] Genes related to chemokine response were downregulated in HCD compared to MCD. This suggests that PMHs experience less stress in HCD conditions compared to MCD, which may contribute to their improved functionality and viability. To further validate the hepatic phenotype of PMHs in HCD and MCD conditions, qRT-PCR was performed on a panel of hepatocyte markers including critical hepatic transcription factors (Hnf4a, Ttr, and Tat), key serum proteins (Alb and Ahsg), serum glycoprotein homeostasis mediator Asgr2, and Serpinf2 which encodes alpha 2-antiplasmin and regulates the blood clotting pathway. The results are provided in FIG. 4C. The expression of all these genes was elevated in HCD PMHs compared to MCD. This elevation indicates better maintenance of hepatic functions in HCD conditions. The expression levels of the fetal hepatic marker AFP were not significantly different between the two conditions, suggesting that PMHs in both groups maintained a mature state.Measuring xenobiotic metabolism in the HCD model

[0069] To further evaluate liver function within HCD constructs, the expression of genes associated with xenobiotic metabolism was analyzed. This analysis included a focus on both phase I and phase II metabolism enzymes as well as hepatic transporters. Heatmap analysis of phase I metabolism genes at day 7 showed significant up-regulation in HCD constructs compared to MCD constructs. Notably, genes related to the cytochrome P450 (CYP) enzyme family such as Cyp3a41 (CYP3A4), Cyp2c55 (CYP2C9), Cyp2bl0 (CYP2B6), and CYPla2 (CYP1A2) were markedly elevated in HCD, indicating enhanced metabolic enzyme functions in PMHs. CYP enzymes are major components of the phase I xenobiotic-metabolism pathway and play a crucial role in the bioactivation of many drugs and contaminants. mRNA expression levels of phase II metabolism enzymes and hepatic transporters were then compared, finding generally higher expression in HCD compared to MCD. Furthermore, gene ontology (GO) analysis revealed that the functional enhancements in HCD were associated with multiple metabolic pathways, including xenobiotic metabolism by cytochrome P450, drug metabolism-cytochrome P450, and retinolmetabolism. Cytochrome P450 pathways are of particular interest due to their crucial role in the metabolism of many medicines and endogenous compounds, underscoring the importance of liver models for drug testing.

[0070] To substantiate these findings, qRT-PCR analysis on the expression of key cytochrome P450 (CYP) enzymes involved in drug metabolism was conducted. Referring to FIG. 5A, the following CYP genes were examined: Cyp3a41, Cyp2c55, Cyp2bl0, Cypla2, Cyp2el, and Cyp7al. These enzymes are collectively responsible for approximately 60% of drug oxidation processes. PMHs in the HCD model exhibited significantly elevated levels of CYP3a41 gene expression compared to the MCD model. Given that CYP3A4 is the most prevalent CYP enzyme and is estimated to be involved in the metabolism of around half of the drugs currently in clinical use, its increased expression underscores the enhanced metabolic capacity of the HCD constructs. The remaining five CYP enzymes also demonstrated higher expression in HCD compared to MCD. The qRT-PCR results confirmed that the HCD liver constructs significantly enhance the expression of major CYP enzymes compared to MCD. This elevated expression aligns with the metabolic functions observed in the RNA-sec results, validating the HCD model as a highly functional and physiologically relevant platform for drug screening and toxicity testing. By ensuring high expression levels of critical CYP enzymes, these constructs provide a robust system for the evaluation of drug metabolism, potentially improving the predictive accuracy of in vitro liver models in pharmaceutical development.Mechanisms of YAP / TAZ pathway regulation in the HCD model

[0071] To further elucidate that the bioprinted HCD model maintains PMH functions in vitro, mechanical cues that regulate the Yes-associated protein (YAP) / transcriptional coactivator with PDZ-binding motif (TAZ) signaling pathway were investigated. Mechanical tension can induce stress fiber formation and subsequent YAP activation, leading to hepatocyte dedifferentiation. Mechanoregulation of the in vitro culture is fundamentally different from the native in vivo microenvironment, significantly impacting cellular morphology and function in short-term cultures. HCD is one main factor affecting the YAP / TAZ pathway. HCD reduces the adhesive area and alters cell shape, leading to inactivation of RhoA and a subsequent reduction in stress fibers within the actin cytoskeleton. This inactivation of YAP / TAZ occurs through both Hippo kinase-dependent and -independent mechanisms. Given that cell-cell interactions have been shown to beenabled in the HCD environment, this suggests that such interactions could play a role in modulating YAP / TAZ activity. FIG. 5B provides a schematic diagram of AJs protein E-cadherin and TJs protein ZO-1.

[0072] Various mechanisms have been proposed to explain how cell-cell contact inhibition modulates YAP / TAZ activities and localization. First, in confluent cells, adherens junctions (AJs) protein E-cadherin trans-dimerizes and stimulates the MST1 / 2-LATS1 / 2 kinase cascade, inhibiting YAP / TAZ activity. Second, tight junction (TJ) protein ZO-1 forms between cells at high confluence, usually involving cis interactions of cell membrane proteins, which also contribute to YAP / TAZ inhibition. Therefore, it was hypothesized that the improved and prolonged hepatocyte function in HCD could be attributed to inactivated YAP / TAZ expression . To test this, the expression of adherens junction (AJ) and tight junction (TJ) genes, specifically Cdhl and Tjp 1 , were evaluated using qRT-PCR (FIG.5C).The expression of these genes was significantly increased in HCD, consistent with previous immunofluorescence staining and flow cytometry results. Next, RNA-seq was employed to examine the YAP-related gene expression profiles in PMHs for both HCD and MCD. Heatmap analysis revealed that classic YAP-related genes were among those down-regulated in HCD. qRT-PCR results confirmed that the expression levels of Yap 1, Wwtrl (TAZ), and their target genes Ctgf and Cyr61 were all significantly lower in HCD (indicative of inactivation) compared to MCD (indicative of activation) (FIG. 5D). These findings demonstrate that the YAP / TAZ pathway is inactivated in HCD, suggesting that improved and prolonged hepatocyte functions in HCD are associated with inactivation of the YAP / TAZ pathway. The increased cell-cell interactions and reduced mechanical stress in HCD contribute to this inactivation, providing a more physiologically relevant environment that supports primary hepatocyte function and viability. This mechanistic insight underscores the importance of cell density and mechanical cues in maintaining functional bioprinted liver models.HCD liver models for drug screening

[0073] In vitro liver models are crucial for drug screening, especially during the early stages of drug development when numerous candidate compounds must be tested. These models need to be high throughput and capable of providing actionable data quickly, typically within 48 hours. To address this need for drug screening applications, the inventive bioprinted liver constructs were placed into a microfluidic chip, creating a dynamic culturesystem that more closely mimics the physiological environment. In addition to assessing the baseline CYP expression levels without any drug treatment, the induction of specific CYP mRNA transcripts was examined by treating the constructs with rifampicin (RIF), a bactericidal antibiotic known for its potential hepatotoxicity risks. Referring to FIG. 6A, the treatment with rifampicin resulted in significant increases in the expression of CYP3 A4, CYP2C9, CYP2B6, and CYP1A2 in HCD. The expression of CYP3A4 and CYP2C9 genes was increased in both HCD and MCD. However, the CYP2B6 and CYP1A2 genes did not show significant upregulation in MCD. Additionally, the expression levels of these CYP genes were markedly elevated even at a lower concentration of rifampicin (20 pM), indicating that PMHs in HCD exhibit heightened sensitivity to this drug treatment. The induction of CYP activity by rifampicin yielded similar results, providing further evidence that PMHs in HCD possess significant potential for drug screening application, as shown in FIG. 6B

[0074] To further evaluate the HCD liver construct’s potential for metabolic drug conversion, diclofenac and midazolam were selected as model compounds due to their well-characterized metabolism by CYP enzymes. Diclofenac is primarily metabolized by CYP2C9, while midazolam is a substrate for CYP3 A4. Monitoring the conversion rates of diclofenac and midazolam will guide the determination of an optimal therapeutic dose, ensuring drug efficacy while preventing toxic side effects during clinical use. After treatment with diclofenac and midazolam for 24 and 48 hours, the drugs and their metabolites in the culture medium were quantified using mass spectrometry (MS). Multiple reaction monitoring (MRM) in negative ion mode was adopted to identify the target fragments, diclofenac (m / z 312.0) and its representative metabolite peaks 4-OH-diclofenac (m / z 230.0), as well as midazolam (m / z 342.0) and its representative metabolite peaks, 1-OH-Midazolam (m / z 203.0) (See Table 3 above).

[0075] FIG. 6C provides standard curves plotting concentration against representative fragment peak area were created using a series of diclofenac and midazolam standards, and the concentrations of diclofenac and midazolam in the samples were calculated based on these linear relationships. The data indicated that MCD constructs exhibited a comparatively lower conversion rate than the HCD constructs (FIG. 6D). The higher conversion rates in HCD constructs suggest an enhanced metabolic capacity, likely due to the improved expression and functionality of CYP enzymes in the HCD environment. Cyclophosphamide(CPA), a widely used oxazaphosphorine prodrug, is easily absorbed but remains inactive until it is metabolized by mixed-function oxidase enzymes (cytochrome P450 system) in the liver. This metabolism yields phosphoramide mustard and acrolein, which alkylate DNA and proteins, respectively. To compare HCD and MCD models and demonstrate the superiority of the HCD model in drug screening, CPA concentrations ranging from 1 pg / mL to 500 pg / mL were employed for a drug toxicity test (FIG. 6E). The toxicity of CPA at MCD was relatively low, even at a concentration of 100 pg / mL, with no dose-dependent effects observed at lower drug concentrations (5-50 pg / mL). However, in the HCD model, cytotoxicity was shown to be concentration-dependent; as the drug concentration increased, increased cell death was observed. Drug toxicity results in static culture showed a similar dose-dependent profile in both HCD and MCD (FIG. 6F). The IC50 results also indicated that the HCD liver model was useful and sensitive in evaluating cytotoxic reactions after the addition of the CPA, as listed in Table 4 below.Table 4IC50 (ug / mL) Dynamic culture Static cultureH 53.543 69.042M 387.257 516.403

[0076] Notably, after 24 hours of CPA treatment, CYP3 A4 expression in HCD cells was found to be dramatically higher than in MCD cells (FIG. 6G). This observation suggests that the PMH in the HCD model successfully converted CPA to its active form, aldophosphamide, via the CYP3 A4 enzyme.Example 2: Synthetic Hydrogel-Based Bioinks

[0077] The second implementation of the bioprinted liver platform evaluated bioinks based on synthetic hydrogels, particularly those based on polyethylene glycol (PEG). Specifically, an 8-arm PEG functionalized with norbornene groups (PEGNB) was created and studied.FIG. 7 diagrammatically illustrates formulation of the PEGNB-based hydrogel, in which PEGNB is crosslinked using an MMP-degradable peptide as the cleavable crosslinker and functionalized with the CYIGSR peptide as a cell-adhesive ligand. This hydrogel incorporates both degradable and non-degradable crosslinks, with their ratio selected to temporally decouple mechanical stability from cellular remodeling. This design allows encapsulated cells to actively remodel the matrix and form mature tight junctions (e.g., ZO-1, E-cadherin), while maintaining the structural integrity of the platform for over 21 or more days. This extended functional lifespan enables the capture of chronic toxicity and metabolic induction responses (e.g., rifampicin-mediated CYP induction) that are typically missed by static or short-term liver models

[0078] Materials: 8-arm polyethylene glycol norbornene (PEGNB, 40 kDa) was purchased from Creative PEGWorks (North Carolina, USA). The peptide CYIGSR, fluorescein-labeled CYIGSR (FAM-CYIGSR), and MMP-sensitive crosslinkers (KCVPMSMRGGCK, MMP-degradable; KCGMMPVSRGCK, scrambled non-degradable) were custom-synthesized by GenScript (New Jersey, USA).

[0079] Cell Culture: iPSC-Heps (iCell Hepatocyte 2.0) were purchased from FUJIFILM Cellular Dynamics. Cells were cultured in hepatocyte culture medium (HCM; CC-3198, Lonza), supplemented with all components from the supplied “SingleQuots” kit and 20 ng / mL oncostatin M (OSM; Sigma-Aldrich), at 37 °C under ambient oxygen and 5% CO2. HUVEC cells were cultured in EGM-2 medium (CC-3162, Lonza).

[0080] Hydrogel Matrix Formulation: Hydrogels were optimized for co-culture of iPSC-Heps and HUVEC. A 3% (w / v) 8-arm PEGNB prepolymer solution was prepared and crosslinked using an MMP-degradable peptide at a 3:1 molar ratio to PEGNB, achieving approximately 75% crosslinking of the PEGNB arms. For iPSC-Heps printing, the formulation also included 100 pg / mL CYIGSR peptide as an adhesive ligand, 30% (v / v) iodixanol, and 0.2% (w / v) lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) as a photoinitiator. For HUVEC printing, both CYIGSR and RGD peptides (cRGDfC) were incorporated at final concentrations of 70 pg / mL and 30 pg / mL, respectively. Non-degradable hydrogels (ND) were formed using exclusively a scrambled crosslinker sequence (KCGMMPVSRGCK), while partially degradable hydrogels (D) combined degradable and non-degradable crosslinkers at a 3:2 molar ratio.

[0081] Bioprinting Process: Three-dimensional (3D) hydrogel scaffolds were fabricated using the BIONOVA X 3D bioprinter (CELLINK, San Diego, USA) under ambient room temperature conditions. Scaffold geometries were digitally designed using computer-aided design (CAD) software with the BIONOVA X slicing platform. The slicing parameters, including layer height, print path, and infill density, were adjusted based on the desired scaffold architecture. Printing was conducted directly onto methacrylated glass coverslips in the well plate, which were functionalized to enable stable anchoring of thephotocrosslinked hydrogel constructs. Photocrosslinking was achieved using a built-in 405 nm UV LED light source at an intensity of 12 mW / cm2, corresponding to 75% of the printer’s maximum output. This controlled exposure facilitated rapid and uniform crosslinking of each printed layer. For multi-cellular printing, each printed zone had an identical thickness. After printing the first layer, excess uncrosslinked hydrogel precursor was gently removed by rinsing with pre-warmed Dulbecco's phosphate-buffered saline (DPBS, 37 °C) to prevent mixing or deformation before printing the subsequent layer. The second zone was then deposited with the same thickness as the first, ensuring consistent structural dimensions throughout the scaffold. Prior to bioprinting, all cells were enzymatically dissociated, passed through 70 pm sterile cell strainers (Corning, NY, USA) to remove aggregates, and centrifuged at 300 x g for 5 minutes to obtain a compact cell pellet. Cell concentrations were determined using a hemocytometer or automated cell counter, and resuspended in the photocrosslinkable hydrogel precursor to the desired final density. For iPSC-Heps bioprinting, the bioink was prepared at a final concentration of 6 x 107cells / mL. For HUVEC bioprinting, a concentration of 3 x 107cells / mL was used.

[0082] Mechanical Testing: Compressive modulus was measured using a MicroSquisher mechanical tester (CellScale). Cylindrical hydrogel pillars (500 pm diameter x 500 pm height) were printed using the same conditions as the experimental liver models. Samples were stored at 37 °C until testing. During testing, stainless steel beams and platens were used to compress the constructs by 10% of their original height over three cycles. Forcedisplacement data from the third cycle were used to calculate compressive modulus using custom MATLAB® scripts.

[0083] Cell Viability Analysis: Live / dead staining was performed on days 1, 3, and 7 postprinting. Samples were rinsed twice with Dulbecco's phosphate-buffered saline (DPBS), then incubated for 30 minutes at room temperature with a staining solution containing 1 pM calcein AM (Cat. #C3099, Invitrogen) and 2 pM ethidium homodimer-1 (Cat. #P3566, Invitrogen). Imaging was conducted using a Leica DMI 6000B fluorescence microscope. Cell metabolic activity was evaluated using the Cell Counting Kit-8 (CCK-8; Cat. #K1018, ApexBio) in quadruplicate. At each time point, samples were incubated in 500 pL culture medium containing 10% CCK-8 for 1 h. 200 pL of supernatant was transferred to a 96-well plate, and absorbance was measured at 450 nm using a Tecan Infinite 200 PRO plate reader.

[0084] Immunofluorescence Staining: For immunostaining, bioprinted constructs were washed three times with DPBS to remove residual media and subsequently fixed in 4% paraformaldehyde for 1 hour at room temperature. After fixation, samples were permeabilized and blocked simultaneously by incubating in a blocking buffer containing 5% (w / v) bovine serum albumin (BSA; SKU#700-101P, GeminiBio) and 0.1% Triton X-100 (Cat. #H5141, Promega) in DPBS. The solution was passed through a 0.22 pm filter to remove particulates before use. Samples were placed on a horizontal shaker and incubated in blocking buffer at 100 rpm for 1 hour at room temperature. Following blocking, constructs were incubated with the primary antibodies (listed in Table 3 above) diluted in Cell Staining Buffer (BioLegend) overnight at 4 °C.

[0085] The next day, samples were rinsed three times with DPBS and incubated with appropriate fluorophore-conjugated secondary antibodies and 4',6-diamidino-2-phenylindole (DAPI, nuclear counterstain) in the dark for 1 hour at room temperature. All antibody incubations were performed on a shaker at 100 rpm to ensure uniform staining. After final rinses in DPBS, samples were imaged using a Leica SP8 confocal fluorescence microscope (Leica Microsystems).

[0086] Cytochrome P450 Activity: The functional activity of Cytochrome P450 enzymes CYP2C9 and CYP3A4 was assessed using the P450-Glo™ CYP2C9 / 3A4 Assay Kit (Promega), following the manufacturer’s protocol. Prior to the assay, printed liver constructs were washed gently with DPBS to remove serum and residual metabolites. Constructs were then incubated with the specific luminogenic substrates provided in the kit for 1 hour at 37 °C under standard culture conditions. After substrate incubation, 25 pL of supernatant from each sample was transferred into a white opaque 96-well plate and mixed with 25 pL of the provided detection reagent. The mixture was incubated in the dark at room temperature for 20 minutes to stabilize the luminescent signal. Luminescence intensity was then measured using a multimode microplate reader with an integration time of 1 second per well. Blank controls containing assay reagent without cells were used to determine the background signal, which was subtracted from all experimental values.

[0087] Quantification of Albumin and Urea Secretion Levels: To assess the hepatic function, albumin secretion and urea production were quantified from culture supernatants collected at defined time points. Collected media were centrifuged at 1,000 * g for 5 minutes to remove any cell debris and stored at -80 °C until analysis. Quantitative measurementswere performed using the QuantiChrom™ BCG Albumin Assay Kit and the QuantiChrom™ Urea Assay Kit (BioAssay Systems), according to the manufacturer’s instructions. Standard curves were generated for each assay to determine absolute concentrations of albumin and urea in the samples. To ensure accuracy and comparability across conditions, all secretion data were normalized to the viable cell population at the corresponding time point, as determined by the CCK-8 assay. This normalization allowed for accurate assessment of per-cell functional output and comparison across different experimental groups.

[0088] RNA Isolation and Quantitative Real-time Reverse-transcription PCR (qRT-PCR):Total RNA was extracted using TRIzol™ reagent (Life Technologies) following the manufacturer’s protocol. To obtain sufficient RNA for analysis, ten bioprinted liver constructs were pooled prior to extraction. Constructs were incubated in TRIzol, and total RNA was purified using the Direct-zol™ RNA Microprep Kit (Cat. #R2060, Zymo Research), which includes a spin-column purification step. Purified RNA was eluted in RNase-free water, and the concentration was quantified using a Tecan microplate reader. RNA samples were immediately stored at -80 °C until further use. For qRT-PCR analysis, 200 ng of total RNA was reverse-transcribed to complementary DNA (cDNA) using the First Strand cDNA Synthesis Kit (Cat. #E6300S, New England BioLabs). Quantitative PCR was then performed using the Luna® Universal qPCR Master Mix (Cat. #M3003S, New England BioLabs) on a QuantStudio™ 3 Real-Time PCR System (Applied Biosystems). Gene expression levels were normalized to housekeeping gene expression using the comparative Ct method (AACt). Forward and reverse primers were synthesized by Integrated DNA Technologies (IDT), and all primer sequences are listed in Table 5.Table 5Primer Primer sequence (5’-3’)CDH1 F GCCTCCTGAAAAGAGAGTGGAAGR TGGCAGTGTCTCTCCAAATCCG TJP1 F GTCCAGAATCTCGGAAAAGTGCCR CTTTCAGCGCACCATACCAACC MMP1 F ATGAAGCAGCCCAGATGTGGAGR TGGTCCACATCTGCTCTTGGCA ALB F GATGAGATGCCTGCTGACTTGC R CACGACAGAGTAATCAGGATGCC HNF4A F GGTGTCCATACGCATCCTTGACR AGCCGCTTGATCTTCCCTGGAT TTR F CGTGCATGTGTTCAGAAAGGCTG R CTCCTCAGTTGTGAGCCCATGC CYP3A4 F CCGAGTGGATTTCCTTCAGCTGR TGCTCGTGGTTTCATAGCCAGC CYP1A2 F TCATCCTGGAGACCTTCCGACAR GCCACTGGTTTACGAAGACACAG CYP2B6 F ACAGTGTGGAGAAGCACCGTGAR GGTTGAGGTTCTGGTGGCTGAA CYP2C9 F CAGAGACGACAAGCACAACCCTR ATGTGGCTCCTGTCTTGCATGC GSTA2 F CTGCCCTTTAGTCAACCTGAGGR ACAAGGTAGTCTTGTCCGTGGC NAT1 F AGGAAGCTCTGGATGACCTTGGR CGTCAAACGGAAGACACAAGGC NAT2 F GGTGGTGTCTCCAGGTCAATCAR TGCAGGAGAAGGTGAACCATGC UGT1A1 F GCAAAGCGCATGGAGACTAAGGR GGTCCTTGTGAAGGCTGGAGAG HIF1A F TATGAGCCAGAAGAACTTTTAGGCR CACCTCTTTTGGCAAGCATCCTG SOD2 F CTGGACAAACCTCAGCCCTAACR AACCTGAGCCTTGGACACCAAC IL-1 F GCCATGTTGTTGTCTGATCGR TGCCATGTGGTGATTTTCTC TNF-a F CTCTTCTGCCTGCTGCACTTTGR ATGGGCTACAGGCTTGTCACTC ACTB F CACCATTGGCAATGAGCGGTTC R AGGTCTTTGCGGATGTCCACGT GAPDH F GTCTCCTCTGACTTCAACAGCG R ACCACCCTGTTGCTGTAGCCAA

[0089] RNA Sequencing and Data Analysis: RNA sequencing (RNA-seq) was conducted to evaluate transcriptomic profiles of bioprinted liver constructs at different conditions: (1) degradable (D) vs non-degradable (ND), (2) co-culture (CO) and mono-culture (mono), and (3) dynamic and static. Total RNA was isolated using TRIzol™ reagent as described above. RNA integrity and concentration were assessed using the Qubit® RNA Assay Kit on theQubit® 3.0 Fluorometer (Thermo Fisher Scientific), and the fragment size distribution of the RNA libraries was evaluated with the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, USA). Libraries were diluted to a final concentration of 1 ng / pL and sequenced on a NovaSeq 6000 S4 platform (Illumina), yielding approximately 40-60 million reads per sample. Sequencing and quality control were performed by Novogen Inc. using three biological replicates per condition. Raw reads were aligned to the Mus musculus reference genome (GRCm39) using HISAT2 v2.0.5. Gene expression levels were quantified as fragments per kilobase of transcript per million mapped reads (FPKM) using HTSeq-count v0.6.0. Differentially expressed genes (DEGs) were identified using DESeq2 vl .20.0, with thresholds set at a fold change > 1 and a false discovery rate (FDR)-adjusted p-value < 0.05. Heatmaps were generated based on log2-transformed fold change (log2-FC) values to visualize DEGs associated with hepatic function and fibrosis. All heatmaps were plotted using the SRplot heatmap tool.

[0090] Microfluidic Liver Chip Culture: A p-Slide III 3D Perfusion chip (Ibidi, Cat. #80376) was used for dynamic culture. Chips were pre-warmed for 1 hour prior to use. A 5 mm coverslip containing the printed liver construct was inserted into the chip and filled with 30 mL HCM. The chamber was sealed with an Ibidi polymer adhesive coverslip. Perfusion was maintained at 0.1 mL / min at 37 °C (Elveflow Cobalt).

[0091] CYP Induction: To evaluate the induction of cytochrome P450 (CYP) enzymes, both gene expression and enzymatic activity levels were assessed in drug-treated and untreated liver constructs. Constructs were treated with drug at two concentrations for 48 hours. Control groups were cultured in medium containing drug. CYP gene expression was analyzed via quantitative real-time PCR (qRT-PCR) as described above, and enzyme activity was quantified using the P450-Glo™ CYP Assay Kit (Promega), following the manufacturer’s protocol. Relative quantification of gene expression was calculated using standard curves and normalized to the expression of the housekeeping gene GAPDH and ACTB.

[0092] Mass Spectrometry (MS) Analysis for Quantification of Cellular Drug Metabolism:To evaluate phenotype-specific drug metabolism, each group was exposed to a cocktail of established cytochrome P450 (CYP) substrates. Following treatment, culture supernatants were collected and mixed with methanol at a 1:2 (v / v) ratio. The mixtures were vortexed thoroughly and centrifuged at 10,000 rpm for 10 minutes to achieve protein precipitationand salt removal. The resulting supernatants were diluted with ultrapure water (1 :8, v / v) and analyzed by high-performance liquid chromatography coupled with tandem mass spectrometry (HPLC-MS / MS) using a Micromass Quattro Ultima triple quadrupole mass spectrometer. Samples (2 pL injection volume) were loaded onto a Cl 8 reverse-phase column, with chromatographic separation carried out at a flow rate of 3 pL / min and a column temperature of 40 °C. Specific metabolites and parent compounds were identified and quantified based on their mass-to-charge (m / z) ratios as listed in Table 6. Metabolic conversion rates were calculated from the concentration of the detected metabolites, enabling comparison of CYP activity across groups.Table 6Target Substrate Metabolite concentration m / zStandard ImipramineCYP2C9 Diclofenac 4-OH-Diclofenac 2 pM 312.0 ^ 230.0 CYP3A4 Midazolam 1-OH-Midazolam 4 pM 342.0 ^ 203.0 CYP2D6 Bufuralol 1-OH Bufuralol 10 pM 278.0 > 186.0 CYP2C19 Mephenytoin 4-OH-Mephenytoin 10 pM 235.1 — > 150.1 CYP1A2 Phenacetin Acetaminophen 10 pM 152.0— > 110.0 CYP2B6 Buproprion OH-Bupropion 10 pM 256.0— > 238.0

[0093] Drug Response Assessment: To determine the drug responsiveness of the liver model, the cocultured model was matured for 18 days and exposed to various types of drugs in static or perfusion condition. Hepatotoxicity testing was performed using non-hepatotoxicants (dexamethasone and aspirin), and hepatotoxicants (acetaminophen (APAP), Troglitazone (TGZ), Voriconazole (VRZ) and Valproate (VLP)). Following drug exposure to the matured liver model, the effects were evaluated by estimating cell viability and lactate dehydrogenase (LDH) activity. Viability was assessed after 24 hours using a CCK-8 test. Absorbance results at a wavelength of 450 nm were recorded for each concentration in each model.

[0094] Determination of the Benchmark Dose (BMP): Cell viability data were quantitatively analyzed using BMD modeling with the PROAST package in R (version 66.20), accessed via the European Food Safety Authority (EFSA) platform and conducted in accordance with EFSA technical guidance. The BMD approach estimates the exposure dose corresponding to a predefined benchmark response (BMR), based on the interpolation of a fitted dose-response curve. In this study, a BMR of 5% was selected, and the BMD wasreported as the lower bound of the two-sided 90% confidence interval (BMDL). Confidence intervals were derived using parametric bootstrap sampling to account for model uncertainty. Finally, BMDL values obtained from the viability assays were used to rank compound potency across different treatment conditions.

[0095] Statistical Analysis: Sample populations were compared using a t-test or one-way ANOVA performed with GraphPad Prism (GraphPad Software). *p < 0.05, **p < 0.01 and ***p < 0.001 were used as the thresholds for statistical significance. Data points on the graphs represent mean values with error bars representing SEM.

[0096] Optimized PEGNB Bioinks Facilitate High-Fidelity Bioprinting and Cell-Mediated Matrix Remodeling: In the present embodiment of the inventive scheme, a synthetic, bioactive, and MMP-degradable hydrogel was developed to recapitulate the physiologically relevant microenvironment of healthy liver tissue. The formulation was based on an 8-arm PEGNB macromer, with each arm having a number-average molecular weight of 5 kDa (total 40 kDa). This molecular weight was chosen to achieve an elastic modulus within the physiological stiffness range of healthy liver tissue (1-7 kPa), which has been shown to support hepatocyte phenotype maintenance in 3D culture.

[0097] The hydrogel network was crosslinked with an MMP-degradable peptide sequence, enabling cell-mediated remodeling, and functionalized with the adhesive peptide CYIGSR to promote hepatocyte adhesion and spreading. Crosslinking and peptide conjugation were carried out through a light-triggered thiol-ene click reaction, a highly efficient and orthogonal photochemical process that proceeds with quantitative conversion under mild, cell-friendly conditions. This reaction enables precise spatial and temporal control over network formation, minimizes side reactions with biological components, and ensures uniform incorporation of bioactive ligands throughout the hydrogel matrix.

[0098] To identify optimal bioink formulations for DLP bioprinting, a range of PEGNB and crosslinker concentrations were tested to assess gelation kinetics, print fidelity, and resolution. Printing outcomes were classified as (i) uncured, (ii) overcured with distorted geometries, and (iii) fully cured with high shape fidelity to the CAD model. A composition of 3% or 5% PEGNB with 75% MMP crosslinker yielded robust, self-supporting constructs with excellent resolution. Although 3% PEGNB with 100% MMP linker produced high-quality prints, this composition left no reactive sites for CYIGSR incorporation, leading to selection of 3% PEGNB with 75% MMP crosslinker for subsequent studies. Cell viabilityassays revealed that CYIGSR concentrations up to 100 pg / mL maintained iPSC-Heps viability above 95% after 7 days (FIG. 8A). Fluorescent labeling with FAM-CYIGSR confirmed its homogeneous distribution throughout the hydrogel and complete coupling via the thiol-ene reaction (FIG.8B). On average, 6-7 PEG arms were used for crosslinking and 1-2 arms for CYIGSR conjugation. To mimic the stiffness of healthy liver tissue, DLP printing parameters (exposure time and light intensity) were fine-tuned to achieve spatially uniform stiffness between 2-6 kPa. This range matches clinically reported values for healthy liver parenchyma. To confirm hydrogel stability, the PEGNB hydrogel was incubated in PBS or hepatocyte culture medium for 3 weeks, showing no significant change in stiffness or thickness. Enzymatic degradation assays using collagenase to mimic MMP activity demonstrated complete degradation of the MMP-sensitive PEGNB hydrogel within 60 mins.

[0099] For long-term dynamic culture, hydrogel stability was first evaluated using high-density hepatocyte encapsulation (6 x 107cells / mL), a density known to enhance hepatic function in 3D culture. However, MMP secretion and flow-induced shear stress caused structural breakdown in fully degradable hydrogels within 3 weeks. To address this, ratios of non-degradable and degradable crosslinker were varied. Hydrogels with >60% degradable linker were found to lose structural integrity after 3 weeks, prompting selection of a ratio of 40% degradable: 60% non-degradable as the degradable (D) condition and 100% non-degradable as the non-degradable (ND) condition for subsequent studies. Importantly, varying the D : ND ratio did not affect CYIGSR distribution or initial matrix stiffness.

[0100] To validate degradability, acellular hydrogels were incubated in collagenase. The D formulation degraded completely within 60 minutes, whereas ND hydrogels remained intact (FIG. 8C). During degradation, the pore size increased markedly (FIG.8D), consistent with previous reports that MMP-driven network cleavage enlarges the mesh size, enhancing the cell-cell contacts and mass transport. Because high cell density reduces hydrogel stiffness, 5% PEGNB with 75% MMP linker was selected for printing hepatocyte-laden constructs. Mechanical testing confirmed that in D hydrogels, the stiffness decreased progressively over 3 weeks, reflecting ongoing cell-mediated degradation, whereas ND hydrogels maintained stiffness over the same period. In both cases, the modulus remained within the physiologically relevant 2-6 kPa range (FIG. 8E). These findings demonstrate thatencapsulated hepatocytes can secrete MMPs to remodel the matrix, enabling dynamic adaptation of the engineered microenvironment.

[0101] Degradable Hydrogels Promote Enhanced Hepatic Organization and Function: The degradable hydrogel (D) formulation was examined to determine whether it could promote the in vitro establishment of physiologically relevant 3D cell-cell and cell-matrix interactions. iPSC-Heps were encapsulated within either degradable (D) or non-degradable (ND) PEGNB hydrogels and evaluated for hepatic phenotype, structural organization, and metabolic competence. The working hypothesis was that in D hydrogels, hepatocytes would actively remodel their surrounding extracellular matrix through MMP-mediated proteolysis, increasing pore size and relieving physical confinement. This remodeling would facilitate tighter intercellular contacts, more extensive cell-ECM adhesion, and long-term network stability — ultimately preserving in vivo-like hepatic morphology and enhancing metabolic performance during prolonged culture (FIG. 9A).

[0102] To assess these organizational differences, immunofluorescence staining was performed for (1) E-cadherin, an epithelial adherens junction protein critical for hepatocyte polarity and survival, (2) ZO-1, a tight-junction marker essential for bile canaliculi formation and maintenance of tissue barrier integrity, and (3) albumin, a major secretory protein that reflects synthetic liver function. Compared with ND, D hydrogels exhibited markedly more continuous E-cadherin and ZO-1 localization along cell borders, indicative of well-formed epithelial junctions, and higher overall albumin intensity. Quantitative image analysis, shown in FIG. 9B, confirmed significantly greater proportions of ZO-1 -positive and E-cadherin-positive cells in D compared to ND, consistent with stronger cell-cell cohesion and epithelial organization.

[0103] Hepatic metabolic activity was further evaluated by measuring the cytochrome P450 (CYP) enzyme activity, which underlies drug metabolism and xenobiotic clearance in vivo. As shown in FIG. 9C, activities of CYP2C9 and CYP3A4 — two major phase I metabolic enzymes — were both significantly elevated in D hydrogels relative to ND, suggesting that the enhanced 3D architecture also supports hepatocyte maturation and metabolic competence.

[0104] Referring to FIG. 9D, anabolic and catabolic performance were assessed by monitoring albumin secretion and urea production over a 3 -week culture period. In both hydrogel types, secretion profiles peaked early and then gradually declined, but D hydrogelsconsistently maintained higher levels across all time points. This sustained functionality in D indicates that the degradable microenvironment better preserves hepatocyte synthetic and nitrogen-metabolism capacities over long-term culture.

[0105] To investigate molecular mechanisms underlying these functional differences, bulk RNA-seq analysis was conducted on iPSC-Heps from D and ND constructs. Principal component analysis (PCA) revealed a pronounced separation between the two groups, confirming global transcriptional divergence. In D hydrogels, the top upregulated Gene Ontology (GO) categories were enriched for metabolic and biosynthetic processes, including cytochrome P450 activity, lipid metabolism, and cell-junction organization, while downregulated genes were predominantly associated with cell-cycle progression and mitotic division. Gene set enrichment analysis corroborated these trends, revealing higher expression of gene networks linked to cell-cell adhesion, ECM organization, and liver-specific functional programs (LiGEP) in D. In addition, GSEA analysis revealed that the degradable environment promoted upregulation of cell-cell junction pathways, such as focal adhesion assembly and cell Junction assembly pathway upregulation.

[0106] qRT-PCR on selected markers further validated these transcriptomic findings. D hydrogels exhibited increased expression of CDH1 (E-cadherin) and TJP1 (ZO-1), consistent with improved intercellular junctions, along with HNF4A, TTR, and ALB levels - markers of critical hepatic transcriptional regulators, transport proteins, and secretory capacity, respectively, shown in FIG. 9E. Notably, MMP-1 expression was higher in ND hydrogels. In healthy adult liver, MMP-1 expression in hepatocytes is minimal, but it rises sharply during chronic liver injury such as non-alcoholic steatohepatitis (NASH), where it is linked to pathological ECM degradation and compromised tissue integrity. This observation suggests that hepatocytes in ND hydrogels experience a stressed microenvironment — characterized by poor ECM remodeling and weaker intercellular junctions — which may predispose them to injury-like phenotypes over time.

[0107] The results presented in FIGs. 9A-9E demonstrate that the degradable (D) hydrogel environment promotes robust hepatocyte network formation, strengthens cell-cell and cellmatrix interactions, and sustains hepatic metabolic function over time compared to the non-degradable (ND) condition. In D hydrogels, MMP -mediated matrix remodeling generates larger pores and improves intercellular connectivity, leading to increased CYP activity, albumin secretion, and urea production. Transcriptomic and gene expressionanalyses further reveal that D hydrogels are enriched for liver-specific functional pathways and junction-related genes, whereas ND hydrogels display transcriptional features indicative of limited ECM remodeling and early hepatic stress.

[0108] Multilayer iPSC-Hepatocyte / Endothelial Co-Culture Mimics Sinusoidal Architecture and Enhances Hepatic Function: The liver acinus, the smallest functional unit of the liver, consists of hepatic parenchymal cells, primarily hepatocytes, and non-parenchymal cells such as liver sinusoidal endothelial cells, organized within a specialized vascular network. Blood from the portal vein and hepatic artery flows through the sinusoidal channels toward the central vein, enabling efficient exchange of oxygen and signaling molecules between endothelial cells and hepatocytes. This intimate hepatocyte-endothelial interface is essential for maintaining hepatic function and metabolic homeostasis.

[0109] To recapitulate key features of this microarchitecture in vitro, a multi-cellular liver construct incorporating iPSC-Heps and HUVECs was designed. While HUVECs are not liver-specific sinusoidal endothelial cells, they can provide important endothelial functions in co-culture, including secretion of angiocrine factors, promotion of hepatocyte polarization, and modulation of metabolic activity via paracrine signaling. The construct was fabricated using a digital light processing (DLP) based bioprinter (BIONOVA X, CELLINK) to achieve precise multi-cellular constructs, positioning endothelial cells in close proximity to hepatocytes. The cell ratio of iPSC-Heps to HUVECs was set at 4:1, consistent with the approximate hepatocyte-to-endothelial ratio in the human liver. By ensuring uniform hepatocyte-endothelial contact and supporting molecular exchange, this engineered geometry more accurately recapitulates the liver sinusoidal environment than organoid cultures or randomly mixed 3D co-cultures (see, e.g., inset 120 in FIG. 1A).

[0110] Performance of the co-culture multi-cellular model (CO) was compared against a hepatocyte monoculture (MONO). Over a 3-week culture period, the CO model consistently outperformed MONO in functional assays. FIG. 10A shows that CO constructs secreted significantly higher levels of albumin, with a slower decline after the initial peak, indicating prolonged synthetic capacity. Urea production was also elevated in CO across all time points, reflecting improved nitrogen metabolism. Referring to FIG. 10B, CYP activity measurements showed that CO (left bar of each pair) maintained higher basal activity of multiple drug-metabolizing enzymes relative to MONO (right bar of each pair), suggestingthat the endothelial component helps sustain hepatocyte metabolic competence over extended culture.[OHl] To explore the molecular basis of these functional improvements, transcriptomic profiling of CO and MONO constructs was performed. PCA revealed clear separation between the two groups, confirming that co-culture induces substantial shifts in gene expression. Gene Ontology (GO) enrichment analysis indicated upregulation of biological processes (BP) related to metabolism and molecular functions (MF) including transcriptional regulation, catalytic activity, and enzyme activity in CO (FIG. 10C). In addition, GSEA analysis suggested that cell junction organization, cell substrate adherens junction assembly and organic acid transmembrance transport are upregulated in CO. Consistent with these findings, KEGG pathway analysis showed activation of multiple interconnected metabolic routes, including xenobiotic metabolism by cytochrome P450, drug metabolism through CYP and non-CYP enzymes, and conjugation / detoxification pathways. Notably, the strong enrichment of cytochrome P450-related pathways underscores the capacity of the co-culture environment to support a complete drug-processing framework, aligning with the improved albumin, urea, and CYP activity observed in functional assays.

[0112] To further characterize the molecular advantages of the CO model, the expression of gene families involved in drug processing, detoxification, and transport was examined. Heatmap analysis at day 14 revealed a distinctive CO expression profile marked by strong induction of multiple cytochrome P450 (CYP) isoforms (CYP3A4, CYP2C9, CYP2B6, CYP2A), conjugation enzymes (GSTA2, NAT1, NAT2, UGT1A1), and various hepatic transporters. This coordinated activation indicates that CO does not simply upregulate isolated metabolic enzymes but re-establishes an integrated drug-handling network encompassing oxidative transformation, conjugation, and export pathways.

[0113] Referring to FIG. 10D, qRT-PCR validation confirmed the RNA-seq findings, showing consistently higher expression of these key enzymes and transporters in CO than in MONO. Together, these transcriptional patterns align with the enhanced albumin production, urea synthesis, and CYP activity observed in the functional assays, underscoring that the CO model supports a broader and more integrated repertoire of hepatic clearance functions.

[0114] Overall, the convergence of functional assays, RNA-seq data, and targeted qRT-PCR validation demonstrates that the multilayer hepatocyte-endothelial co-culture model markedly improves liver-specific gene expression and functional performance compared with monoculture. By supporting high levels of key CYP enzymes, phase II enzymes, and transporters, the CO model represents a physiologically relevant and metabolically active platform for drug metabolism and toxicity studies, offering improved predictive accuracy over conventional hepatocyte monocultures.

[0115] Dynamic Perfusion Preserves Hepatic Metabolic Function by Relieving Hypoxia, Reducing ROS, and Maintaining Mitochondrial Activity: Although FIGs. 9A-9E and 10A-10D demonstrated that both the MMP-degradable matrix and hepatocyte-HUVEC co-culture system promote cell-cell and cell-matrix interactions to enhance metabolic performance, these benefits plateaued and then declined under static culture conditions. Specifically, albumin secretion, urea production, and CYP activity peaked around day 14, after which all measures progressively decreased. This decline reflects a well-documented limitation in long-term static cultures: the inability to maintain an optimal microenvironment for metabolically active hepatocytes. To address this, a dynamic perfusion culture system was introduced. Previous studies have shown that dynamic flow platforms enable continuous transport of nutrients, metabolites, and oxygen while preventing the accumulation of waste products. This is particularly important for metabolically active liver tissues, where oxygen depletion is a hallmark of deteriorating culture conditions and can trigger dedifferentiation and inflammatory signaling. In the inventive high-cell-density model, static culture exacerbates this challenge, as oxygen diffusion becomes rate-limiting. Hypoxic stress, in turn, promotes oxidative stress and inflammation, both of which impair hepatic metabolic function.

[0116] FIGs. 11A-11F provide results of comparing the degradable co-culture liver model in dynamic versus static culture, revealed that dynamic perfusion preserved albumin secretion, urea production, and CYP activity throughout the full 3 -week culture period, with no significant decline from the day-14 peak. In FIGs. HA- 11B, static culture exhibited a sharp decrease after day 14, confirming that dynamic flow extends the functional lifespan of the in vitro liver model. ROS quantification (FIG. 11C) revealed no difference between static and dynamic groups during the first week. However, by weeks 2 and 3, ROS levels increased sharply in static culture, even with daily medium changes. In contrast, ROS levelsremained low in dynamic culture, demonstrating that perfusion effectively mitigates oxidative stress by maintaining oxygen supply to relieve hypoxia and by removing metabolic by-products.

[0117] The mechanism by which dynamic culture maintains metabolic function was next examined. Hypoxic stress not only increases ROS but also impairs hepatic antioxidant capacity and disrupts mitochondrial activity. This is particularly critical in high-density cultures, where oxygen diffusion limitations are more severe, leading to metabolic zone heterogeneity reminiscent of ischemic injury in vivo. Transcriptomic analysis revealed that dynamic culture preserved three key aspects of mitochondrial function: ribosome integrity (supporting mitochondrial protein synthesis and respiratory enzyme turnover), matrix function (facilitating substrate oxidation and TCA cycle flux), and oxidative phosphorylation capacity (enabling high-yield ATP production under prolonged culture). Together, these findings indicate that perfusion preserves mitochondrial ribosome integrity, matrix function, and oxidative phosphorylation capacity, thereby preventing the ROS-driven membrane and mtDNA damage that compromises respiratory chain function in static culture. GSEA analysis at day 21 showed significant enrichment in the dynamic culture group for: (i) oxidative stress detox and redox regulation and (ii) mitochondrial and energy metabolism. KEGG pathway analysis further demonstrated that dynamic culture, by alleviating oxidative stress and preserving mitochondrial function, leading to the upregulation of multiple metabolic pathways compared to static culture. This enrichment pattern underscores the synergistic effect of sustained oxygen delivery and efficient waste removal, which limit ROS accumulation and protect mitochondrial integrity. In turn, healthy mitochondria maintain high-capacity energy production and redox balance, creating a positive feedback loop that supports long-term hepatic metabolic activity.

[0118] qRT-PCR validation confirmed that oxidative stress-responsive genes (HIF1A, SOD2) were less induced in dynamic culture (left bar in each pair) compared to static (right bar) at day 21 (FIG. HD). This lower induction is consistent with reduced hypoxic signaling and lower mitochondrial ROS load. Notably, SOD2, the mitochondrial superoxide dismutase, functions as a primary defense against superoxide radicals generated in the electron transport chain. Its relatively modest upregulation in dynamic culture suggests that mitochondria remained in a healthier, less oxidatively stressed state, rather than being forced into a compensatory antioxidant response. Referring to FIG. HE, inflammatory markers(TNFa, IL-ip) showed a slight increase in static culture at day 21, but the difference was not statistically significant, likely because inflammatory processes occur preferentially in hypoxic cellular sub-regions but are not solely dependent on hypoxia phenotypes. In parallel, metabolic function-associated genes, including key liver markers and CYP isoforms, shown in FIG. 11F, were consistently higher in dynamic culture at day 21. This molecular evidence complements the functional data shown in FIGs. 11A-11B, showing that dynamic perfusion not only sustains but also molecularly reinforces hepatocyte metabolic capacity over extended culture periods. In summary, dynamic perfusion culture effectively disrupts the hypoxia-ROS-mitochondrial damage cycle that limits the performance of static high-density hepatic cultures. By continuously delivering oxygen and nutrients, removing waste products, and sustaining mitochondrial integrity, dynamic culture prevents oxidative damage, maintains antioxidant capacity, and preserves the transcriptional programs underlying hepatic metabolic function. These combined effects extend both the functional lifespan and the metabolic competence of engineered liver tissue well beyond the limits of conventional static culture.

[0119] Dynamic Bioprinted Liver Models Demonstrate Enhanced Metabolic Competence and Clinically Relevant Drug Response: Bioprinted liver models offer a promising platform for high-throughput drug screening, particularly in the early phases of drug development when rapid evaluation of numerous compounds is required. To demonstrate the metabolic competency of the inventive model, a panel of compounds processed by specific cytochrome P450 (CYP) enzymes, whose metabolic pathways are well characterized, was selected. Monitoring the conversion rates of these substrates into their respective metabolites provides a quantitative measure of enzymatic activity. Analysis of metabolite formation serves as a key indicator of hepatic function and supports dose optimization to enhance therapeutic efficacy while minimizing potential toxicity. After a 48-hour treatment with the selected substrates, drug conversion and metabolite formation were quantified using mass spectrometry (MS). Parameters for LC-MS / MS conversion rate quantification of drug metabolites are provided in Table 6 above.

[0120] As shown in FIG. 12A, the inventive liver model maintained under dynamic culture conditions exhibited significantly higher conversion rates compared to static culture, particularly for substrates metabolized by CYP3A4, CYP2C9, CYP2B6, and CYP1A2. In contrast, conversion rates for CYP2C19 and CYP2D6 substrates showed no substantialdifference between the two conditions. These enhanced conversion rates in dynamic culture suggest improved hepatic metabolic function, consistent with elevated expression of metabolism-related genes and pathways observed in earlier analyses (FIGs. 11A, 11B, and 11F). To further examine the metabolic competency of the model, CYP gene induction and enzymatic activity was evaluated following treatment with specific CYP inducers. As shown in FIG. 12B, dynamic culture significantly upregulated the mRNA expression of CYP3 A4, CYP2C9, CYP2B6, and CYP1A2. Correspondingly, CYP enzyme activities measured in FIG. 12C showed corresponding increases, confirming that the bioprinted liver model under dynamic conditions exhibits inducible drug-metabolizing capabilities. These results collectively underscore the platform’s potential for high-throughput drug screening.

[0121] Following a 21-day post-bioprinting maturation period, the bioprinted liver constructs were exposed to two well-characterized non-hepatotoxic compounds — aspirin and dexamethasone — under both static and dynamic culture conditions. Drug exposures were conducted for 24 and 48 hours to assess tissue response under different durations and conditions. As shown in FIG. 12D, aspirin exposure at clinically relevant concentrations resulted in comparable metabolic activity between the two culture conditions, with no notable variation in drug-metabolizing function. Cell viability remained above 95%, and intracellular lactate dehydrogenase (LDH) activity was preserved across all doses tested within the therapeutic range, indicating maintenance of hepatocellular integrity and low cytotoxic burden. However, when aspirin was administered at concentrations exceeding 1000 pM, both static and perfused liver constructs exhibited signs of cytotoxicity. This was evidenced by a marked reduction in DNA content and a concomitant decrease in LDH activity, suggesting impaired cell viability and compromised cellular membranes at supratherapeutic doses. These observations align with previously reported clinical outcomes and FDA data, which document the hepatotoxic potential of high-dose aspirin. The ability of the bioprinted liver model to recapitulate this clinically relevant toxicity profile underscores its predictive utility for drug safety assessment. In contrast, referring to FIG.12E, treatment with dexamethasone across the selected concentration ranges did not adversely affect tissue viability or hepatic function. No significant changes were observed in DNA content or LDH release after 24 and 48 hours of exposure in either culture condition. The absence of cytotoxic or metabolic impairment following dexamethasone treatment further validates the physiological fidelity of the model. This result is consistent with itsclinical safety profile and highlights the model’s ability to distinguish hepatotoxic and non-hepatotoxic compounds, reinforcing its utility as a predictive in vitro platform for drug screening and hepatotoxicity assessment.

[0122] Dynamic 3D Bioprinted Liver Model Improves Predictive Sensitivity for Clinically Relevant Drug-Induced Hepatotoxicity: Liver toxicity is a leading cause of late-stage clinical trial failure and post-market drug withdrawal, largely due to the limited predictive power of traditional animal models and static in vitro liver systems. These models often fail to capture the complexity of human liver physiology, especially in detecting human-specific toxic responses. To address this gap, a physiomimetic in vitro liver model was developed using multi-cellular 3D bioprinting of iPSC-derived hepatocytes and endothelial cells within a degradable matrix. The construct is integrated with a microfluidic platform to enable dynamic perfusion, enhancing tissue maturation, function, and cell-cell communication. This design mimics key features of the liver microenvironment, including spatial organization and mechanical stimuli.

[0123] The drug screening potential of the inventive 3D bioprinted liver model was evaluated using four DILIrank compounds with established hepatotoxicity profiles: acetaminophen (FIG. 13A), valproate (FIG. 13B), voriconazole (FIG. 13C), and troglitazone (FIG. 13D). The model was cultured under both static and perfused conditions to compare responses in different microenvironments. After 48 hours of drug exposure, both culture systems exhibited dose-dependent toxicity, as indicated by decreased DNA content, elevated LDH release, and reduced cellular viability. Notably, at higher drug concentrations, acetaminophen (5000 pM), voriconazole (50 pM), and troglitazone (100 pM) induced significantly greater reductions in cell viability and LDH activity. The perfused model showed enhanced sensitivity to drug-induced toxicity compared to static condition, with a more pronounced loss in viability and metabolic activity. These findings indicate that dynamic perfusion promotes more uniform drug distribution and more accurately reflects physiological responses, aligning with prior 3D liver model studies that report superior metabolic competency and toxicological sensitivity.

[0124] To further evaluate the predictive accuracy of the inventive bioprinted liver model for drug-induced hepatotoxicity, the system’s half-maximal inhibitory concentration (ICso) and benchmark dose (BMD) values for a panel of hepatotoxic compounds with their respective human maximum plasma concentrations (Cmax) were compared. As shown inFIG. 13E, regression analysis revealed a stronger correlation between Cmax and both ICso and BMD values in the dynamically cultured liver constructs compared to the static culture condition. This suggests that the dynamic environment, which mimics physiological fluid flow, enhances the functional maturity and metabolic responsiveness of the liver tissue, thereby improving the predictive performance of the model. To further assess the influence of culture duration and dynamic conditions on drug sensitivity, relative potency factors (RPFs) were calculated by normalizing ICso and BMD values obtained from week 3 (W3) mature liver tissues against the corresponding values from week 1 (Wl) tissues as a baseline. The W3 dynamic culture consistently demonstrated increased drug sensitivity, as indicated by lower ICso and BMD values relative to Wl . Furthermore, many of the ICso and BMD values in dynamic culture fell within or below the “>lxCmax” or “>10><Cmax” threshold categories — commonly used benchmarks for clinically relevant toxicity prediction — whereas static cultures more frequently exceeded the “>100><Cmax” range. These findings underscore the enhanced physiological relevance and utility of the dynamic culture system in detecting clinically meaningful hepatotoxic response.

[0125] The development of predictive human in vitro drug screening platforms requires integration of advanced biomaterials, precise architectural control, and physiologically relevant culture conditions. In the foregoing description of the second embodiment, these needs were addressed by engineering a multi-cellular liver construct built from an MMP-degradable PEGNB hydrogel and matured under dynamic microfluidic perfusion. Together, these design elements recapitulate essential structural, biochemical, and mechanical cues of the native hepatic sinusoid, resulting in long-term maintenance of liver-specific function and clinically relevant drug-response profiles.

[0126] Central to this strategy is the MMP-degradable PEGNB hydrogel, which combines biochemical customizability with rapid, uniform crosslinking via thiol-ene photopolymerization. This chemistry is particularly advantageous for DLP-based 3D bioprinting, as it supports high-speed, high-fidelity fabrication while preserving the viability of encapsulated cells. The modular PEGNB platform allows independent tuning of stiffness, degradability, and ligand presentation, enabling us to closely mimic the compliant, cell-interactive environment of healthy liver tissue. By incorporating an MMP-cleavable peptide crosslinker, a remodelable hydrogel network that permits encapsulated hepatocytes to locally degrade and reorganize their extracellular matrix was created. This remodelingprocess, which is absent in non-degradable matrices, provides a key for maintaining hepatocyte morphology, polarity, and intercellular junction integrity.

[0127] The benefits of a degradable matrix extend beyond structural mimicry. MMP-mediated remodeling promoted cell-cell contact formation, reinforced epithelial junctions, and facilitated bile canaliculi-like structures, all of which are closely tied to hepatic function. Functionally, degradable hydrogels supported higher albumin secretion, urea synthesis, and cytochrome P450 activity over prolonged culture compared to their non-degradable counterparts. Transcriptomic analysis confirmed enrichment of pathways associated with cell junction organization, focal adhesion assembly, and lipid metabolism, alongside reduced expression of stress-associated genes such as MMP1, which is often elevated in injury-like microenvironments. These observations reinforce the concept that ECM degradability is not merely a passive structural property but an active regulator of hepatocyte phenotype stability and metabolic competence.

[0128] To further recapitulate the liver’s microanatomy, multi-cellular bioprinting was employed to organize iPSC-Heps and endothelial cells into distinct yet closely apposed zones. This spatial arrangement mimics the sinusoidal structure of the liver, where hepatocytes are positioned adjacent to endothelial channels to facilitate nutrient exchange and paracrine communication. Such architectural fidelity cannot be achieved with randomly mixed co-cultures or self-organized organoids, which, although capable of some degree of cellular self-organization, lack the reproducibility and directional organization needed for consistent metabolic output. In mixed cultures, endothelial-parenchymal interactions are stochastic and unevenly distributed, leading to functional heterogeneity. By contrast, the inventive multi-cellular design ensures uniform endothelial support signals throughout the construct, producing sustained hepatocyte maturity, elevated basal and inducible CYP activity, and more predictable drug metabolism. This underscores a key advantage of 3D bioprinting: the ability to precisely dictate cellular composition and spatial arrangement in a reproducible, scalable manner.

[0129] While multi-cellular co-culture provides the blueprint for a physiologically relevant liver model, dynamic microfluidic perfusion supplies the necessary microenvironmental cues for long-term functional stability. Static cultures, even with optimal architecture, eventually suffer from diffusion limitations, leading to oxygen depletion, nutrient gradients, and waste accumulation. These factors drive localized hypoxia, which triggers oxidativedi¬stress, mitochondrial dysfunction, and phenotypic deterioration. In the multi-cellular model disclosed herein, dynamic perfusion is able to effectively mitigate these issues by continuously delivering oxygen and nutrients while removing metabolic by-products. This prevented the hypoxia-ROS-mitochondrial damage cascade that typically limits the lifespan of static liver cultures. As a result, perfused constructs-maintained albumin and urea secretion, CYP activity, and mitochondrial integrity for over three weeks without the functional decline observed in static conditions.

[0130] The mechanism underlying the preservation of function involves protection of mitochondrial health and suppression of oxidative stress. Dynamic culture maintained mitochondrial ribosome integrity, preserved TCA cycle enzyme expression, and sustained oxidative phosphorylation capacity. ROS levels remained low, and antioxidant enzyme induction (e.g., SOD2) was modest, reflecting a low oxidative burden compared to static culture, where antioxidant pathways were strongly upregulated in response to accumulated stress. Interestingly, no significant increase in inflammatory gene expression was observed in dynamic culture. This may be explained by the fact that inflammatory processes in dense hepatic constructs arise preferentially in hypoxic micro-regions, where ROS accumulation and damage-associated molecular patterns stimulate immune-like signaling. By preventing the formation of these hypoxic niches, perfusion reduced the triggers for inflammation. Furthermore, as inflammation is not solely dependent on hypoxia, the absence of alternative pro-inflammatory cues in the inventive well-balanced biochemical and structural microenvironment likely further contributed to the stable, low-inflammatory state.

[0131] The combined effects of degradable ECM, multi-cellular sinusoidal organization, and dynamic perfusion yield a platform that is not only structurally and functionally robust but also predictively accurate in drug response testing. In drug screening assays (Figures 6-7), the system distinguished non-hepatotoxic compounds (aspirin, dexamethasone) from clinically recognized hepatotoxicants (acetaminophen, troglitazone, voriconazole, valproate) with high fidelity. Perfused constructs displayed enhanced sensitivity to toxic compounds, reflected in significantly lower ICso and benchmark dose (BMD) values compared to static culture. Crucially, these potency metrics aligned closely with reported human plasma maximum concentration (Cmax) thresholds for hepatotoxicity, often falling within or below clinically relevant ranges (>lxCmax or >10xCmax). Regression analysisdemonstrated stronger correlations between Cmax and both ICso and BMD values for dynamic cultures, underscoring the translational relevance of the inventive system.

[0132] This high predictive accuracy likely stems from the model’s ability to replicate the physiological determinants of drug metabolism and toxicity — including sustained CYP activity, preserved mitochondrial function, and realistic drug distribution dynamics under perfusion. Moreover, the reproducibility of bioprinting ensures consistent construct geometry and cell distribution, minimizing batch-to-batch variation that can obscure drug response signals in more stochastic culture systems such as organoids. Together, these attributes position the multi-cellular, perfused, and degradable liver platform as a scalable, high-content, and human-relevant alternative to conventional animal models for preclinical drug evaluation.

[0133] The inventive platform has diverse commercial applications in pharmaceutical development and personalized medicine including: 1. Drug metabolism and toxicity testing, accelerating preclinical testing by providing a reliable and scalable platform; 2. Personalized drug screening, supporting tailored drug testing using patient-derived iPSCs, improving therapeutic efficacy and safety; 3. Hepatotoxicity studies to enable precise evaluation of drug-induced liver injury (DILI); and 4. Disease modeling, providing insights into liver disease mechanisms and treatment development.

Claims

CLAIMS:

1. A multicellular liver model comprising:a construct 3D bioprinted from a bioink to form a high cellular density (HCD) liver acinus structure, wherein induced pluripotent stem cell (iPSC)-derived hepatocytes and supporting cells are incorporated into the construct during printing.

2. The multicellular liver model of claim 1, wherein the supporting cells are one or a combination of liver sinusoidal endothelial cells (LSECs), hepatic stellate cells (HSCs), human umbilical vein endothelial cells (HUVECs), cholangiocytes, and immune cells.

3. The multicellular liver model of claim 2, wherein the construct is bioprinted in a plurality of layers comprising:a first layer corresponding to a hepatic chord structure and comprising iPSC-derived hepatocytes; anda second layer corresponding to a vascular layer and having a permeable interface configured for molecular exchange between a nutrient medium and the first layer.

4. The multicellular liver model of claim 3, further comprising an inter-channel formed between the first and second layers, the inter-channel configured to create a metabolic gradient in a fluid flow through the construct.

5. The multicellular liver model of claim 3, wherein the first layer is printed with a cell density within a range of 1 million cells / mL to 1,000 million cells / mL.

6. The multicellular liver model of claim 3, wherein the first layer further comprises cholangiocytes.

7. The multicellular liver model of claim 3, wherein the second layer comprises one or more of LSECs, HSCs, and HUVECs.

8. The multicellular liver model of claim 1, wherein the construct is bioprinted in one or more wells of a multi-well plate.

9. The multicellular liver model of claim 1, where the bioink comprises a mixture of hydrogel and a refractive index matching agent.

10. The multicellular liver model of claim 9, wherein the hydrogel comprises GelMA.

11. The multicellular liver model of claim 9, wherein the hydrogel comprises polyethylene glycol (PEG) functionalized with norbornene groups (PEGNB).

12. The multicellular liver model of claim 11, wherein the PEGNB is crosslinked using an MMP-degradable peptide as the cleavable crosslinker and functionalized with the CYIGSR peptide as a cell-adhesive ligand.

13. The multicellular liver model of claim 12, wherein a ratio of a non-degradable to a degradable crosslinker is selected to decouple mechanical stability from cellular remodeling.

14. The multicellular liver model of claim 13, wherein the ratio is 40% degradable: 60% non-degradable.

15. The multicellular liver model of claim 1, wherein the iPSC-derived hepatocytes are patient-specific.

16. The multicellular liver model of claim 1, wherein the construct is integrated with a microfluidic chamber, the microfluidic chamber configured for in-perfusion of oxygen and nutrients and out-perfusion of metabolic by-products through the liver acinus.

17. The multicellular liver model of claim 16, wherein the microfluidic chamber comprises an inlet and an outlet, the inlet configured for introducing a sample to be screened.

18. The multicellular liver model of claim 16, wherein a plurality of integrated constructs and microfluidic chambers are disposed within individual wells of a multi-well platform, wherein each microfluidic chamber is individually perfused.

19. A method for modeling liver disease and hepatotoxicity comprising: introducing a sample to be screened to the multicellular liver model of claim 16; and monitoring the multicellular liver model for changes in expression of one or more genes indicative of reaction to the sample.

20. The method of claim 19, further comprising inducing different hepatic phenotypes using co-cultures of supporting cells.

Images