A Comprehensive Review on 3D Tissue Models: Biofabrication Technologies and Preclinical Applications

Abstract

The limitations of traditional two-dimensional (2D) cultures and animal testing, when it comes to precisely foreseeing the toxicity and clinical effectiveness of potential drug candidates, have resulted in a notable increase in the rate of failure during the process of drug discovery and development. Three-dimensional (3D) in-vitro models have arisen as substitute platforms with the capacity to accurately depict in-vivo conditions and increasing the predictivity of clinical effects and toxicity of drug candidates. It has been found that 3D models can accurately represent complex tissue structure of human body and can be used for a wide range of disease modeling purposes. Recently, substantial progress in biomedicine, materials and engineering have been made to fabricate various 3D in-vitro models, which have been exhibited better disease progression predictivity and drug effects than convention models, suggesting a promising direction in pharmaceutics. This comprehensive review highlights the recent developments in 3D in-vitro tissue models for preclinical applications including drug screening and disease modeling targeting multiple organs and tissues, like liver, bone, gastrointestinal tract, kidney, heart, brain, and cartilage. We discuss current strategies for fabricating 3D models for specific organs with their strengths and pitfalls. We expand future considerations for establishing a physiologically-relevant microenvironment for growing 3D models and also provide readers with a perspective on intellectual property, industry, and regulatory landscape.

1. Introduction

Drug discovery remains a lengthy and costly process due to its low success ratio during clinical trials, where at least 75% of the novel drugs, which have been demonstrated their efficacy during preclinical testing, failed in the clinical phase because of the insufficient efficacy and relatively poor safety performance [1]. Despite significant progress made within the pharmaceutical sector, a substantial rate of failure continues to be the primary cause of the extensive time and expenses involved in pharmaceutical research [2]. The majority of the total cost and the time taken for the drug discovery cycle are attributed to attrition in drug due to the low predictivity of drug effects and toxicity in preclinical research. To improve research and development output, a paradigm called “quick-win, fast-fail” has been gradually considered by the pharmaceutical community to reduce the attrition rate, ultimately saving costs and time [3]. Increasing predictivity of preclinical testing has been a major focus in this community. Certainly, greater emphasis has been placed on enhancing the success rates in drug development, especially to prevent the poor candidates from the preclinical stage flowing into clinical trials, and the key is to construct new drug targets, which are more precise preclinical models that can accurately simulate in-vivo features, especially microenvironmental factors [4].

Currently, two-dimensional (2D) culture models, where cells proliferate on a flat, rigid plastic substrates designed for enhanced adhesion and proliferation of cells, have been accepted as the standard for drug screening before in-vivo experimentation owing to their cost-effectiveness and streamlined processes [5]. However, growing evidence makes it clear that 2D models often fail to accurately predict drug efficacy in vivo. The main reason for this is that 2D models fail to replicate the intricate microenvironment found in vivo [6]. Indeed, considerable differences could be observed when cells were cultured in monolayer with respect to those in vivo, where the cells are usually surrounded by the extracellular matrix (ECM) secreted by themselves, allowing for interactions between cell to cell and between cell and the ECM that are limited in 2D. These interactions usually mediate the cell morphology, behavior, migration, adhesion, and gene expression [4, 7]. It is not surprising that the 2D models can not accurately reflect the response within the scope of drug testing. Three-dimensional (3D) models, where cells are aggregated as spheroids or embedded within specific matrices before culture, have been making rapid progress recently [8]. Compared to 2D models, interactions between cells and interactions between cells and ECM are strongly found in 3D models. Therefore, some key factors of in-vivo environment can be recapitulated. This makes 3D models an increasingly reliable tool for pharmacological studies, especially drug screening, pharmacokinetic and pharmacological analysis [9, 10]. An ideal 3D model should also include the gradients of pH, oxygen, nutrient, and metabolic waste, which make the model more accurate in the physiological or pathophysiological microenvironment [6, 11]. Table 1 provides a comprehensive comparison between 2D and 3D models, highlighting their distinct characteristics and features.

Table 1.

A comparison of the strengths and weaknesses of 2D versus 3D models

Type of cell culture	Strengths	Weaknesses
2D model	Commercially available	Cells only adhere to and tile on a single surface
	High reproducibility and performance	Lack of cell to cell and cell to matrix interactions
	Ease-to-use protocols	Partial polarization due to the sheet-like cell morphology
3D models	Closely resembling the in-vivo conditions	Expensive and time-consuming
	Provides interaction between cell to cell and cell to matrix like in-vivo	Reproducibility and performance are relatively limited
	in-vivo-like distribution of oxygen, nutrients, and metabolites to cells	The culture procedures are complicated

During previous decades, significant advancements have been achieved in fabricating different types of 3D in-vitro models for preclinical applications, such as disease modeling and drug screening. These advancements have unveiled new opportunities for their use in the field of pharmaceutics. This paper reviews the latest advancements in 3D in-vitro tissue models used for drug screening and disease modeling for variety of organs and tissues, such as the liver, gastrointestinal tract, kidney, heart, brain, bone, and cartilage as highlighted in Figure 1. We explore the current biofabrication technologies to create 3D models for specific organs, discussing their advantages and limitations. Moreover, this is the first review, which covers multiple preclinical applications together taking examples of several tissue types and highlights long-run prospects of building up tissue microenvironments to upgrade the development of 3D models.

Figure 1.

Overall biofabrication pathway for 3D tissue models using different cell sources to target organ specific disease modeling and drug screening.

2. Biofabrication Technologies for Building 3D Tissue Models

The benefits of 3D models over 2D have been widely realized, which accelerated the development of technologies for biofabrication of 3D models. This section highlights different biofabrication technologies for developing 3D models including heterocellular spheroids, hydrogel-based scaffolds, microfluidics, organoids, bioreactors, and 3D bioprinted tissues, as illustrated in Figure 2.

Figure 2.

The common biofabrication technologies for building 3D tissue models

2.1. Spheroid Models

Spheroids are clusters of cells with strong cell-cell adhesion, resembling a small section of tissue, and can be fabricated through various techniques. In the past, spheroid models were grown using the hanging drop technique, a method that allows cells to form into clusters first, and then biologically aggregate in a small suspension [12, 13]. By changing the substrate, the technique allows spheroids to be grown in the usual test tubes, glasses, and orifice plates. In the realm of evaluating drugs, the main problem of the hanging drop technique is that its limited cell culture volume complicates the process of drug introduction, particularly when it involves essential medium changes, which limits its use in high-throughput applications [14, 15]. As an alternative, spheroids can be produced in greater culture volumes by employing orifice plates coated with a non-reactive substrate inhibit cell adhesion to plastic surfaces. Low adherent plates promote cell-cell interactions, fostering the formation of spheroidal structures [16]. The technique is commonly employed to investigate drug responses in tumor spheroids [17, 18]. It is simple, inexpensive, reproducible, compatible with multiple tumor cell lines, and suitable for drug screening on a scale ranging from medium to high throughput. Nonetheless, its flexibility is limited because spheroids float freely in the medium, which hurdles to the process of medium change. It is important to note that some cell lines do not naturally develop spheroidal structures in low attachment environments because of their restricted cadherin expression.

Magnetic levitation is another method of generating spheroids. This technique uses magnetic nanoparticles to provide magnetic environment and cells are incubated with these nanoparticles to form spheroids [19, 20]. This technique can quickly facilitate the formation of spheroids, but its drawback is that the use of nanoparticles may affect cell viability and function in subsequent drug screening experiments. There are other techniques that can be used for the formation of spheroids, such as but not limited to microwell arrays and microfluidic-based technique. The reader is referred to literature reviews on spheroid biofabrication techniques for more information [21].

2.2. Hydrogel-based scaffolds

As interactions between ECM and cells and how such interactions influence cell morphology and biology are especially important for drug screening, this prompted researchers to utilize ECM-like biomaterials for biofabrication of 3D models. Among them, the most common method is cell encapsulation within hydrogels with characteristics similar to the native ECM. According to the material sources of hydrogel preparation, there are two categories: natural and synthetic hydrogels.

2.2.1. Natural Hydrogels and Derivatives

Collagen, Matrigel, fibrin, gelatin, and sodium alginate are the most common natural sources used for the preparation of hydrogels. These natural hydrogels are primarily characterized by their exceptional biocompatibility and cell-friendly properties [22–26], which can support many of the physiological functions of cells.

Collagen, especially Type I collagen, is widely used ECM protein in 3D culture platforms [27]. By changing the concentration and gelation temperature of collagen, its chemical and physical properties can be altered, and then the proliferation and drug response of cells can be modulated [28]. Collagen and Matrigel (isolated from the ECM of mouse chondrosarcoma containing several naturally occurring cytokines and growth factors) modulate cell reaction to treatments, including chemotherapy, immunotherapy, and radiotherapy. For example, when comparing drugs for breast cancer, prostate cancer, and lung cancer, different dose-responsive curves could be observed for Docetaxel and Fulvestrant [29], particularly, when cells were cultured in Matrigel versus other platforms such as collagen and alginate. Collagen and Matrigel solutions are maintained as a liquid at low temperatures. Thus, cell encapsulation is required to be processed at low temperatures, which makes collagen and Matrigel unsuitable for high-throughput drug screening applications when using common room temperature equipment.

Fibrin, a naturally occurring polymer produced during the healing process of wounds, is made by polymerizing fibrinogen from plasma proteins. At present, it has found extensive use in 3D culture applications, including angiogenesis [30] and biomechanics research [31], and mesenchymal stem cell (MSC) culture [32]. However, due to the high sensitivity to protease-mediated degradation, fibrin is challenging to maintain in long-term culture. Therefore, its applications in fabrication of 3D models for drug screening is limited.

Gelatin is a polymer, which is a result of partial chemical and thermal degradation of collagen and can be easily soluble in water at 37 °C. Due to similarity with collagen and its non-immunogenic properties, gelatin finds versatile use in multiple applications, including contact lenses manufacturing, drug delivery and tissue engineering [33]. Chemical properties can be easily modified by the introduction of specific functional groups [34]. Despite of these advantages, fast enzymatic degradation sensitivity, poor mechanical properties, and the limited solubility of gelatin in concentrated aqueous media restricts its applications in drug screening [35].

Another natural substance commonly used to make hydrogels is sodium alginate, derived from the cellular structure of brown algae’s outer layer. Due to the absence of cellular attachment ligands, the application of alginate in 3D culture models has been hampered, but rapid crosslinking property of alginate has led to its extensive application in 3D bioprinting research [13, 36, 37].

2.2.2. Synthetic Hydrogels

The advantage of synthetic hydrogels for 3D models with respect to natural hydrogels lies in all aspects of the performance, thus making synthetic hydrogels a promising alternative [38]. The synthetic hydrogels used to prepare 3D cell culture models mainly include polyethylene glycol (PEG), polyacrylic acid, polyurethane-based hydrogels, and poly (vinyl alcohol) (PVA) [39]. Compared to natural hydrogel, these synthetic hydrogels lack the cell adhesion sites, so the corresponding cell adhesion sites, such as RGD polypeptides, are usually introduced during the preparation process. Synthetic hydrogels have found extensive applications in 3D models for drug screening [40, 41], such as liver models, heart models (co-culture of fibroblasts and beating cardiomyocytes), neuron models (astrocytes and microglia), etc. In these models, synthetic hydrogels can be used as scaffolds for the preparation of tissue models, enabling in-vitro toxicological experiments and high-throughput drug screening.

2.3. Emerging biofabrication technologies

Microfluidic-based micro-physiological systems or organ-on-chip technologies have been considered advanced platforms for fabrication of 3D models. These technologies can control interaction between cells and interactions between ECM and cells and enable perfusion capabilities, but adapting cell-cell and cell-ECM interactions to microplates for high-throughput screening tools is challenging. Conversely, 3D bioprinting has surfaced as a technology for creating reproducible and high-throughput 3D models, with the capability of incorporating tunable ECM components to replicate in-vivo conditions. These characteristics make the technique particularly suitable for drug screening.

2.3.1. Organ-on-a-chip platforms

Organ-on-a-chip platforms are specifically engineered for cell and tissue culture under continuous perfusion, ensuring a consistent delivery of oxygen and nutrients while effectively eliminating waste products [42]. They provide the advantage of accurate manipulation of the tissue microenvironment, enabling a consistent and regulated release of growth factors and nutrients. In addition, it is particularly suitable for studying cell-cell interactions among various cell types. For example, microfluidic control has been employed for studying the impacts of tumor necrosis factor apoptosis-inducible ligand and curcumin on prostate cancer cells, where cells cultured in microfluidic systems exhibited different IC₅₀ values after a combinatorial treatment compared to cells cultured in 96-well plates, suggesting that the microfluidic system was more effective than the well-plate system in evaluating cell-based combination treatments [43]. Kwapiszewska and colleagues designed a platform made up of microcells and micropores and placed human liver and colon cancer cells behind the microcells to form spheroids, and then monitored the metabolic activities of these spheroids for two weeks. In addition, cell pellets were treated with various concentrations of 5-Fluorouracil over a period of 12 days, and cell viability assay showed that cell pellets in microfluidic devices showed higher concentrations of resistance compared to those in 2D [44].

2.3.2. Organoids

Organoids are originally referred to the isolated stem cells from the body and then cultured in natural hydrogels (such as Matrigel) to gradually develop into aggregated complex 3D structures with similar functions of the corresponding in-vivo organs [45]. Until now, organoids have been grown in dishes. Presently, organoid cultures encounter challenges, including the absence of a natural microenvironment (e.g., growth factor gradient, ECM composition) and the absence of communication with immune cells to replicate immune responses. Transcriptome analysis is a widely used downstream organoids application; however, its capacity for drug discovery and targeted therapy requires additional development and exploration. For example, organoids have found application in toxicity screening studies to assess the impacts of Cisplatin [46] and drug screening in organoid models of cystic fibrosis has also been reported [47]. In addition, tumor cells and the microenvironment derived from patient tumors offer an enhanced and more accurate 3D culture system for personalized drug evaluation and advancement. The pairing of tumor organoids with adjacent healthy tissues can lay a foundation for constructing organoid biobank as a resource pool for drug screening and development [48].

2.3.3. 3D Bioprinting

3D Bioprinting is an advanced biofabrication technology that enables the creation of intricate structures with exceptional precision, especially to restore functions of tissues in vivo [49]. Using specific bioinks, positioning of cells can be precisely finely tuned to closely resemble the tissue microenvironment more realistically. Recent advancements in bioink compositions and crosslinking techniques, and process techniques have broadened the utilization of 3D bioprinting in preclinical applications [50, 51]. Notably, 3D bioprinting substantially enhances the effectiveness of producing 3D methods. The predominant 3D bioprinting technologies encompass extrusion-based, laser-assisted, light-based, and droplet-based bioprinting.

The dominant approach in bioprinting is extrusion-based bioprinting (EBB), which has also been employed for producing 3D tissue models used in drug screening and toxicology investigations [52–54]. It enables the construction of 3D tissue formations sequentially, layer by layer via dispensing cell-laden bioinks through a nozzle. The relatively low resolution of the bioprinted tissues, and inherent limitations, such as cell viability being compromised by shear stress during the extrusion process are some of its drawbacks [55]. Laser-assisted bioprinting (LAB) utilizes laser pulses laser pulses for transferring bioink from donor slides to receiver slides to generate 3D tissue models. LAB enables the deposition of individual cells in a single droplet at a high resolution, which makes it highly unique in the realm of high-resolution applications [56]. The poor scalability and high cost of LAB limit its practical value for generating of 3D models for high-throughput screening purposes. Light-based bioprinting (LBB) involves the use of ultraviolet or visible light for curing a photosensitive bioink, enabling the step-by-step construction of highly precise models for the creation of complex biological tissues [57]. Stereolithography, a type of LBB, stands out from other techniques because of its capacity to rapidly produce tissues with exceptional precision, which may cause negative effects to cell viability and function [49, 58]. Newer technologies such as digital light processing has been adapted to facilitate the high-throughput fabrication of tissue in well plates [59]. Droplet-based bioprinting (DBB) evolved from conventional commercial inkjet printers and have been employed for printing bioinks containing cells for the fabrication of 3D tissue models [60]. Cell-laden bioinks are sprayed from multiple nozzles via heat or piezoelectric. By modifying the voltage, one can regulate the size and form of bioink droplets within the picoliter range which enables DBB to accurately deposit cells to a specific site, ensuring a highly biogenic model is constructed. Bioprinted cells also typically have high viability, bringing DBB an incomparable advantage when dealing with precious cell samples, especially those obtained from patients [61]. The major advantages of DBB are accuracy, non-contact nature of deposition and high-throughput ability making it a favorable option for applications that require rapid fabrication of tissues. However, DBB has some weaknesses. It uses high pressure to generate droplets that can damage cells and decrease their viability, potentially affecting the functionality of tissue models. Furthermore, bioprinting of complex structures such as blood vessels can be challenging due to the collapsing nature of droplets at the time of landing. Additionally, there may be variability in droplet size and placement, leading to inconsistent and suboptimal bioprinted structures. While DBB has its limitations, it remains a promising technology with the potential to enable high-throughput production of tissue models [62]. In addition to these conventional bioprinting modalities for cell deposition purposes, there are modalities enabling deposition of spheroids or organoids for fabrication of tissue models [63, 64]. For a more extensive understanding of bioprinting processes, readers are directed to recent literature reviews [65, 66]; however, we here summarize very recent tissue or cancer models fabricated by 3D bioprinting for drug discovery in Table 2. Depending on the target tissue or organ type, Figure 3 highlights the strengths and weaknesses of the biofabrication strategies discussed in this section.

Table 2.

Summary of recent 3D research systems fabricated by 3D bioprinting for drug discovery

Bioprinting technologies	Bioink formations	Tissue or cancer model	Drug(s)	Comments	Reference
EBB	Hydrogel formation: BB BM and AA silk fibroin, liver ECM-derived microparticles, gelatin, and with or without β-D galactose. Cells: HLCs, HUVECs, HHSCs	Liver model	Aspirin and Dexamethasone; Trovafloxacin mesylate; Acetaminophen and Troglitazone	A robust platform for hepatotoxicity screening	[67]
EBB	Hydrogel formation: Sodium alginate, gelatin, fibrinogen and aprotinin Cell: H9C2	Cardiac tissue	Digoxin	The effects of digoxin on the sodium ion channel were particularly estimated to prove the potential of drug screen and discovery	[68]
EBB	Hydrogel formation: GelMA and HAMA Cells: HOS, 143B or U2-OS	Osteosarcoma model	Cisplatin, Doxorubicin, Methotrexate, Erastin, Apatinib, Sorafenib, Everolimus, and Chloroquine	Multi-omics analysis revealed the important role of autophagy in osteosarcoma	[69]
EBB	Hydrogel formation: GelMA, CSMA, and HAMA Cells: C4–2B, PC-3 and fibroblast	Prostate cancer model	Hyaluronic acid (HA)	HA induced PCCs to activate CAF transformation and form HA-CAF coupling to promote PCCs drug resistance	[70]
EBB	Hydrogel formation: Fibrinogen and novogel Cells: NDFs, pericytes, endothelial cells derived from iPSCs, NHEKs	Vascularized skin model for studying atopic dermatitis	Dexamethasone; Three Janus Kinase inhibitors: ruxolitinib, baricitnib and tofacitinib	Offers an approach to understand atopic dermatitis relevant disease and platform for evaluating the effectiveness and potential harmful impacts of drugs	[71]
EBB	Hydrogel formation: Fibrin, alginate, genipin, and chitosan Cell: U87	Glioblastoma tumor model	Forskolin, ISX9, CHIR99021, IBET 151, and DAPT.	A glioblastoma model with exceptional accuracy for drug screening when compared to 2D culture.	[72]
EBB	Hydrogel formation: Sodium alginate and gelatin Cell: MCF-7	Breast cancer model	Camptothecin and Paclitaxel	A simple and direct approach for culturing and characterizing drug-resistant spheroids to support anticancer drug screening.	[73]
EBB	Hydrogel formation: Sodium alginate, gelatin, diethyl amino ethyl cellulose, and collagen peptide Cell: A549	Lung cancer model	Mitoxantrone	A hybrid alginate-gelatin-based bioink was optimized and bioprinted as lung cancer model for drug screening application	[74]
EBB	Hydrogel formation: Sodium alginate and gelatin Cells: SW480, THP-1, and HUVECs	Colorectal cancer model	Fluorouracil (5-FU), oxaliplatin, and irinotecan	Expression of tumor-related genes significantly improved more resistant to chemotherapy.	[75]
Contactless micro-jet printing	Hydrogel formation: GelMA and Fibrinogen Cells: HUVECs, ADSCs or iPS-MSC, STA-NB15 cells	Neuroblastoma tumor model	Angiogenic growth factor VEGF-165; Bortezomib	A microvascularized tumor-on-a-chip model with potential for studying the tumor angiogenesis and metastasis	[76]
Digital light processing	Hydrogel formation: PEGDA and GelMA Cells: HepG2 cells and HUVECs	Liver hepatocellular carcinoma model	Doxorubicin	Rapid and direct generation of 3D models within multiwell cell culture plates	[59]
Droplet (microval ve) based bioprintin g	Hydrogel formation: Matrigel Cell: Human muscle precursor cell	Skeletal muscle model	Caffeine and Tirasemtiv	A potential microphysiologi cal system to benefit the development and screen of drugs against muscle wasting diseases	[23]
Digital light processing	Hydrogel formation: GelMA Cells: cholangiocyt es, HepG2, HUVEC;	Cholangiocarcin oma model	Cyclophosphamide	Mimicking native cholangiocarcin oma conditions represents a potential screening method	[77]

Note: BM: Bombyx mori; AA: Antheraea assamensis; HLC: Hepatocyte-like cell; HUVEC: Human umbilical vein endothelial cell; HHSC: Human hepatic stellate cell; GelMA: Gelatin-methacrylate; STA-NB15 cells: patient-derived high-stage neuroblastoma tumor cells; ADSCs: Adipocyte-derived mesenchymal stem cells; H9C2 cells: rat embryonic cardiomyocytes; iPSCs: induced-pluripotent stem cells; HAMA: Hyaluronic acid methacrylate; HOS, 143B and U2-OS :human osteosarcoma cell lines; CSMA: Chondroitin sulfate methacrylate; C4–2B, PC-3: Prostate cancer cell lines; CAF: Cancer-associate fibroblast; PEGDA: Poly(ethylene glycol) diacrylate; HepG2: hepatocellular carcinoma cell line; NDFs: Neonatal human dermal fibroblasts; NHEKs: Neonatal human epidermal keratinocytes; U87: Human glioblastoma multiforme cell lines; ISX9: Isoxazole 9; SW480: Human colorectal adenocarcin9oma cell line; THP-1: Human acute monocytic leukemia cell line; MCF-7: Human breast cancer cell line; A549: Adenocarcinomic human alveolar basal epithelial cell line.

Figure 3.

Strength and weaknesses of biofabrication technologies for building 3D tissue models.

3. 3D Models utilized in Screening Drugs and Modeling Diseases for Various Tissues and Organs

3.1. The Liver

The liver, being the primary organ responsible for maintaining homeostasis, holds a significant role in the metabolism of amino acids, lipids, heterologous metabolism, proteins, glucose, clotting factors, and bile production [78]. In the past decade, numerous research endeavors have focused on development of liver models, which can replicate the functions of the liver. These models are intended for investigating both liver disease-related mechanisms and the impact of drugs [79]. Primary human hepatocytes (PHHs) in 2D culture is the classic model for the evaluations of drug effects and toxicities. However, this model suffers from quickly losing hepatocytic functions. Not only this, but other liver cell lines are also limited by incomplete functions and lack of microenvironments during in-vivo studies. 3D Hepatocyte models offer great promise for studying liver diseases and evaluating drug metabolism and pharmacokinetics (PK). A series of different 3D liver disease models have been developed for liver function studies, which can be used for acute and chronic drug hepatotoxicity prediction, liver fibrosis, fatty liver, liver cirrhosis, and liver cancer drug screening. 3D Models provide more accurate insights into the mechanisms of liver disease, making them a better platform for drug screening [80–84].

3.1.1. 3D Models for Drug Screening for the Liver

Hepatocellular spheroids are the most common liver tissue model. As a scaffold-free mode, this model is simple and low-cost. Compared to 2D culture, it can prolong the culture time while maintaining the viability and phenotype of hepatocytes. Therefore, it is suitable for acute or chronic large-scale evaluation of drug effects on liver functions. For example, Bell et al. reported that PHH spheroids cultured on plates designed for minimal attachment exhibited a resemblance to the liver physiology and even these PHH spheroids maintained their inter-individual variability even throughout a comprehensive proteome analysis. Additionally, they could be sustained under serum-free conditions and can maintain metabolic activity for a minimum of five weeks in culture, supporting versatile models for chronic toxicity assays, cholestatic and steatotic and virus infected disease models [85]. PHH spheroids, when cocultured with nonparenchymal cells, showed a better phenotype, higher secretion of albumin (ALB) and apolipoprotein B (ApoB), along with gene expression associated with lipid metabolism, stage I metabolism, and glucose [86]. These studies suggested that the PHH spheroid system shows great promise as a versatile and highly effective in-vitro model suitable for investigating liver diseases, liver function, drug testing, and the long-term effects of drug-induced liver injury (DILI). Recent research has revealed that expandable human hepatoprogenitor cell-like spheroids (IHEPLPCS-3D) exhibited strong liver-specific function and markers compared to PHH [87]. Besides, since parenchymal cells are significantly instrumental in liver injury, heterocellular spheroids of hepatocytes, Kupffer cells, sinusoidal endothelial cells, and hepatic stellate cells should be considered for liver physiological and pathological models [88, 89]. Similarly, hepatocyte (HepG2) and hepatic stellate (LX-2) spheroid models have shown their potential in simulating physiological and pathological liver conditions [90, 91]. Vorrink et al. used prolonged spheroids culture of human primary hepatocytes serve as an in-vitro model for evaluating drug-induced liver toxicity. The model effectively differentiated between hepatotoxic and non-toxic drugs, demonstrating enhanced sensitivity and specificity [92]. Takayama et al. used human embryonic stem cell (hESC)/iPSC-based hepatoid spheroids to detect 22 chemicals and separate drugs with and without hepatotoxicity. In the screening of hepatotoxic drugs (Acetaminophen and Trioglitazone), they found that IC₅₀ was lower than that of HepG2 spheroids and close to that of PHHs spheroids. These results suggested that hESC/iPSC-based hepatoid spheroids were more sensitive than HepG2 spheroids in predicting drug-induced hepatotoxicity [93]. Recently, an integrated bionic array chip (iBAC) was developed to fabricate hepatocellular spheroids of collagen-coated PHHs. A large-scale liver toxicity screening was conducted using 122 drugs with known clinical toxicity classification. The results showed that the iBAC-based PHH spheroid model had a better prediction of hepatocellular toxicity drugs with the highest sensitivity at 71% [82].

Hydrogels with high biocompatibility and tunability (permeability, elasticity, hardness, and chemical reactivity) can simulate the native tissue microenvironment by maintaining both temporal and spatial aspects of biochemical and physical signals [94]. As one of the 3D culture technologies, hydrogels loaded with liver cells are widely used in drug screening studies. To improve the culture conditions of rat hepatocytes in vitro, a phenylalanine/glutamyl/lysyl self-assembled peptide hydrogel (FEK-SAPH) was developed for pharmacokinetic screening to assess drug uptake, disposal, toxicity (ADMET), excretion, and metabolism in hepatocytes [95]. Luo et al. utilized a nanofiber-based hydrogel to facilitate the conversion of human iPSCs into cell types resembling hepatocytes, which enhanced the effect of hepatocellular enzyme inducers on CYP450 enzyme activity and can be used for drug screening [96]. HepG2 cells grown on hyaluronic acid (HA), named HA³P50, proliferated into larger aggregates of cells with controlled release of liver cell-specific bioactive substances (bile acids, albumin, urea, transaminases), along with the synthesis of cytochrome P450(CYP)7A1 [97].

Liver-on-a-chip devices have been used for growing liver tissue under continuous perfusion to simulate the physiological functions of the liver, which can show dose and time-dependent hepatotoxicity after exposure to drugs [98]. For example, Corrado et al. cultured HepG2 in a microfluidic device with three parallel channels and a shape like that of hepatic sinuses. To evaluate detoxification potential of the device, samples were treated with different concentrations and treatment duration of ethanol. The findings demonstrated that albumin and urea exhibited a decrease that was dependent on the dosage [99]. Another study used a liver-on-a-chip platform designed for efficient drug toxicity screening on a large scale. iPSC-derived liver cell (iHep) aggregates were loaded in ECM within the organ channel, which was isolated from the vascular channel of 96 two-channel microfluidic chips. Endothelial cells and macrophages derived from THP-1 monocytes were cultured together in the vascular channel of the microfluidic chip. A total of 159 compounds known for their effects on the liver were added to the constructed model to calculate toxicological priority scores, dose-response assessments using standard laboratory liquid handling robots to seed, deliver, retrieve, and supplement media and substances [80].

The 3D bioprinting technology has also been applied in the development of intricate liver models in vitro [100]. Compared to traditional 2D monolayer culture, bioprinted hepatocytes showed enhanced liver-specific functionality and drug metabolism capacity after cytochrome P450 induction [101]. Bioprinted HepaRG and hepatoid models with human hepatic stellate cells showed higher expression of ALB and CYP3A4 compared to monolayer culture of HepaRG [102]. In a study, researchers used a bioink comprising of PHHs and collagen-HA to liver models that maintains urea and albumin production for up to 2 weeks and responds to drugs accordingly [103]. Overall, the use of bioprinting to construct in-vitro liver models shows significant advantages for enabling long-term culture while preserving liver specific function and drug metabolism capabilities in an automated manner.

3.1.2. 3D Disease Models for the Liver

Animal models remain widely utilized models for the preclinical evaluation of liver fibrosis. However, drugs with potential effects in animal models of fibrosis have not yielded successful results in clinicals, indicating disparities between animals and humans [104]. A representative and robust in-vitro model of human hepatic fibrosis is urgently needed. In recent times, significant advancements have occurred in the progress of bioengineering of liver fibrosis models, which simulate the pathophysiological progression of liver fibrosis and provide appropriate in-vitro models for evaluating drugs with anti-fibrotic properties [105]. HepaRG co-cultured with human stellate cells are suitable for conducting repetitive assessments of drugs targeting liver fibrosis and shows different toxicity and activation characteristics of hepatic stellate cells according to drugs properties. In this regard, Leite et al. used low-adhesion 96-well plates to form HepaRG/HSC spheroids, which exhibited hepatic fibrosis characteristics, such as activation of hepatic stellate cells (HSCs), secretion of collagen, and its subsequent deposition after 14 days of single or multiple exposures to pro-fibrotic compounds (allyl and methotrexate) [90]. Researchers induced HSC cells (iPSC-HSCs) in vitro and gathered HepaRG cells into spheroids to construct a model of liver fibrosis and subjected them to transforming growth factor-beta (TGF-β) and thiocaracetamide (TAA). iPSC-HSC/HepaRG and primary HSC/HepaRG spheroids showed the same levels of expression of fibrosis markers and the secretion of type I collagen precursors. An advancement in liver fibrosis characteristics such as increased phalloidin and collagen staining was observed, suggesting that iPSC-HSC/HepaRG could serve as a promising approach in identifying novel therapeutic targets for liver fibrosis [106]. A spheroid model consisting of HepaRG, HUVECs, and LX-2 cells that were stimulated was also utilized for liver fibrosis. This model showed decreased CYPs metabolic activity, decreased albumin production, and up-regulation of ECM-related genes. When Galunisertib was added, it was discovered that the genes (Collagen1A2, ACTA2) and α-smooth muscle actin protein (α-SMA) responsible for producing fibers decreased in a manner that was influenced by the dose [107]. In a separate investigation, Norona and colleagues utilized EBB for the construction of a 3D liver microstructures consisting of parenchymal (primary human liver cells) and non-parenchymal (HSC and endothelial cells) cells, which were exposed to different pro-fibrosis drugs and held promise to be used as a high-throughput model for pre-clinical anti-fibrosis drug screening [108]. Overall, 3D liver fibrosis has been extensively utilized as versatile tools for disease modeling objectives.

Fatty liver is commonly divided into two categories: non-alcoholic fatty liver disease (NAFLD) and alcoholic fatty liver disease (AFLD). AFLD resulting from over drinking stands as the primary chronic liver, while NAFLD is among the rapidly developing diseases [109]. Currently, the common in-vitro model for the fatty liver is the 2D culture of hepatocytes. Research has shown that the 2D model cannot mimic the in-vivo injury effectively [110]. Thus, there is a need to develop a sophisticated 3D in-vitro model of fatty liver for the purpose of studying the disease mechanism and drug response. In a study, Jiang and colleagues showcased the development of in-vitro alcoholic liver disease (ALD) model, where HepG2 cells were encapsulated within a dual hydrogel scaffold made of HA and collagen, and the disease was induced with alcohol. Thereafter, lipogene, expression was increased, which indicates the successful establishment of an alcoholic fatty liver model. Along with HepG2, other cell types including LX-2, human umbilical vein cell line (EAhy926) and KC (U937) cells were utilized to constitute a demountable liver-on-a-chip model under dynamic perfusion to study the pathophysiology of alcohol-induced single non-parenchymal cells in AFLD and the outcomes indicated that this model replicated the complex arrangement and microenvironment of hepatic sinusoids and has potential to recreate the damage induced by alcohol [111]. Spheroids have also been used for developing fatty liver models. After adding Sorafenib (15μM) for 72h, expression of TGFB1, COL1A1, PDGF, and TIMP1 genes was down-regulated, and the protein levels of IL-6, TNF-α, and TGF-β1 were decreased, and the steatosis induced fibrosis was also inhibited [112]. These results indicate that Sorafenib has an anti-fibrosis effect on steatosis-induced fibrosis model of NAFLD. Pingitore et al. used a heterocellular spheroid model composed of hepatic stellate cells (LX-2) and hepatocytes (HepG2) to investigate new compounds during high-throughput drug screening for nonalcoholic steatohepatitis (NASH). While Liraglutide and Irabenol (experimental drugs for NASH) were added, the symptoms of NASH were relieved [91]. In another study, Duriez et al. co-cultured PHH, HSC, LEC and Kupffer cells on collagen and added free fatty acids and TNF-α to the medium to establish a 3D NASH model, which maintained cell activity for 10 days. This model was adequate for medium-scale throughput anti-NASH drug screening and provided a desirable approach to gain a deeper comprehension of the physiological aspects of the disease and to recognize and characterize novel drug targets more effectively [113]. Using the microfluidic chip technology, Bulutoglu et al. established a gradient structure to distribute the medium containing linoleic acid by concentration to explore the heterogeneity of pathogenesis and physiological structure of liver tissue in NAFLD [114]. To study the pathogenesis and conducting drug screening for NAFLD, a bioprinted liver tissue was also used to recapitulate fatty liver in vitro. However, at present, the development of NAFLD models mostly stays in the stage of steatohepatitis and liver fibrosis, and it is still a challenge to develop models related to liver cirrhosis and liver cancer, which requires a multidisciplinary approach to promote their development.

Hepatocellular carcinoma (HCC) ranks among the most prevalent tumors in the world [115]. In this regard, researchers used bioprinting to successfully build an HCC model using a bioink composed of primary liver patients loaded in gelatin/alginate. When four anti-liver cancer targeting drugs, namely Levastinib, Sorafenib, Regogfinib and Apatinib, were added, the results showed that hepatocellular carcinoma cells of different patients were resistant to different drugs, suggesting that the HCC model serves as a reliable model for a long-term in-vitro culture and can predict specific drugs for personalized treatment [116]. In a different research investigation, Wang and colleagues constructed a 3D endothelialized liver tumor model, comprised of porous microspheres based on poly(lactic-co-glycolic acid) (PLGA-PMs) loaded with human liver cancer cells and HUVECs (Figure 4). The IC₅₀ of PLGA-PMS to anticancer drugs (including Cisplatin and Doxorubicin) was higher than that of the 2D model [117]. Based on the modular construction strategy of PLGA-PMs, various cell types could be chosen as parenchymal and non-parenchymal cells to replicate the targeted tumor microenvironment. This model has potential significance for in-vitro antitumor drug screening and provides a favorable platform for cancer treatment. Further, Ma et al. employed 3D bioprinting to create a liver tissue model using dECM with customizable mechanical properties. This model was designed to facilitate the study of HCC progression, and it stably recapitulated the clinically relevant mechanical properties of cirrhotic liver tissue through 3D bioprinted liver dECM scaffolds. In cirrhotic-like conditions, reduced growth and increased invasion markers were observed in HepG2 cells seeded within the dECM scaffolds compared to healthy controls. This work offers a promising platform for in vitro disease modeling with tunable mechanical properties [118].

Figure 4.

3D Endothelialized liver tumor model to screen anticancer drugs (including Doxorubicin (DOX) and Cisplatin (CIS)). A) In the model, HepG2 cells and HUVECs were co-cultured, as illustrated in a schematic diagram. B) Under dynamic culture conditions, DOX was taken up by HepG2 cells and HUVEC aggregates. C) Scanning electron microscope (SEM) images to analyze the arrangement of sizes and shapes of PLGA -PMs. D) The fabricated PLGA-PMs were subjected to analysis for pore size distribution and Fourier transform infrared (FTIR). E) Cytotoxicity tests comparing the cell viability following incubation with varying concentrations of DOX and CIS. F) Fluorescence microscopy images depicting the AO/EB staining of cells grown in both 2D as well as 3D environments, taken at different time durations (2, 4, and 12 h) (reproduced/adapted from Ref. [117]).

3.2. The Gastrointestinal (GI) Tract

The GI tract refers to the digestive tract ranging from the stomach to the anus, including gallbladder, small intestine, large intestine, and rectum. It is the main tract of the digestive system and crucial for the process of food digestion, absorption of nutrients, immune protection, and the synthesis of hormones [119]. The high incidence of GI diseases remains a significant issue affecting public health. There is a growing body of research to address this health issue, and preclinical drug testing depends on data derived from in-vitro models [4]. In general, cells are cultured in 2D but recent studies for implementing 3D methods have shown valuable insights [120]. 3D GI models for evaluating drugs and personalized therapeutic disease modeling are gradually being established.

3.2.1. 3D Models for Drug Screening for the GI Tract

Spheroid models are commonly used in drug screening and development for the GI tract. Samy et al. reported geometrically-regulated self-assembled multicellular spheroids of Caco-2 cells resembling the intestine showed a better representation than 2D models, demonstrated by Caco-2 cells differentiating into intestinal epithelial phenotypes more rapidly and surviving longer in culture (Figures 5A–B) [121]. This intestinal model offers a potential platform to expedite drug screening processes and study the function of intestinal transporter proteins. In another study, intestinal epithelial cells differentiated from human jejunal spheroids was used as a model to understand how drug-metabolizing enzymes and transporter proteins affect drug absorption in the GI tract [122]. These models enable a quantitative evaluation of how enzymes responsible for metabolizing drugs in the intestines and the proteins that facilitate their transport impact the uptake CYP3A substrate drugs in the intestine.

Figure 5.

3D Drug screening models for the GI tract. A) A schematic illustration depicting the creation and application of agarose micro-molds to create patterns of Caco-2 spheroids within Matrigel. B) At Day 21, the spheroids retained their viability and lumen structure, as demonstrated by the LIVE/DEAD assay depicting their maintained viability after being cultured for 21 days (reproduced/adapted from Ref. [121].) C) Microscopic pictures of HT29-MTX-E12 tumor spheroids after exposure to nanoparticles for 3 h, under both static and dynamic conditions of HT29-MTX-E12 tumor spheroids in the presence of Lumogen^® F Red 305-loaded nanoparticles, either attached to the spheroids surface or permeated within them (reproduced/adapted from Ref. [123]).

3D Models show differences in morphology, proliferation rates, and drug response under static conditions with responses and functions significantly amplified under flow conditions [124, 125]. For instance, Elberskirch and colleagues established a microfluidic intestinal model highlighting a surface characterized by a mucus layer and microvilli [123]. The setup involved fluid flow, enabling the assessment of functionalized nanoparticles capability to adsorb and enter the mucus layer under active conditions. For model testing, two distinct systems consisting of particles at the nanoscale (surface-coated with Carbopol^® or Pluronic^® F127) were formulated to assess their capacity for adsorption and penetration through the mucus layer. The results demonstrated both nanoparticles without coatings and those with surface coatings had significant adsorption and permeation properties in both static and dynamic systems. The microfluidic intestinal model serves as a suitable preclinical platform for screening new drugs (Figure 5C).

Currently, most GI models rely on Matrigel, which provides the necessary ECM elements like fibronectin, collagen, and laminin [126, 127]. In another study, on the other hand, it was demonstrated that alginate supported the growth of the human intestinal organoid (HIO) in vitro and promoted the differentiation of HIO epithelial that closely resembled the differentiation observed in HIOs cultured in Matrigel. Moreover, when transplanted in vivo, the HIOs cultured in alginate reached a comparable level of maturation as the HIOs grown in Matrigel. The pure mechanical support provided by an easily accessible and cost-effective hydrogel is adequate to support the survival and growth of HIOs [128]. Decellularized ECM (dECM) from the GI tract was also used as an alternative to Matrigel for culturing GI tract-like organoids suitable for drug screening [129].

Benefits of 3D bioprinting has become a significant driver in the rapid advancement of GI models for drug screening. For example, a human intestinal villi model was developed using Caco-2-cells, where villi structures were achieved by controlling the composition of bioink (dECM and collagen) as well as the bioprinting parameters [130]. In-vitro cellular activity has shown that collagen/dECM were more relevant in mimicking the layer of epithelial cells that comprises the intestinal structure compared to the villus structure composed solely of cell-laden collagen, which can be applicable for drug screening and advancement purposes.

3.2.2. 3D Disease Models for the GI tract

Along with spheroids or hydrogels, organoids generated from human pluripotent stem cells (hPSCs) or adult stem cells (ASCs) [131], have recently become popular models for disease modeling [132]. For example, researchers developed hPSC-derived HIOs to utilize as Crohn’s disease fibrosis model via TGF-β induction [133]. For the TGF-β treatment, HIOs were enclosed within growth factor-reduced Matrigel to test the anti-fibrotic drug Spironolactone. They showed that the Spironolactone treatment blocked fibrosis caused by TGF-β induction. In another study, Boj et al. used trichothecene-induced intestinal organoid swelling to assess in-vitro drug response in individuals diagnosed with cystic fibrosis [134]. The swelling caused by Trichothecene was entirely reliant on the cystic fibrosis transmembrane conductance regulator (CFTR). The method exhibited sensitivity and accurately differentiated drug responses among individuals, regardless of whether they had different or identical CFTR mutations. This method offers a cost-efficient approach to identify drug-responsive individuals. Hale et al. used iPSC-derived organoids and macrophages to study host-bacterial interactions, including enterotoxigenic Escherichia coli (ETEC) and Vibrio cholerae [135]. In a separate study, Kane and colleagues employed human iPSC-derived small intestine tissues for modeling ETEC-producing and Vibrio cholera-induced diseases for drug screening and detection, and personalized treatment of these diseases [136].

For GI tumors, as transformed cell lines fail to represent the complexity and physiology of native tumors and differences in development and physiological characteristics exist between human and animal models, patient-derived organoids have emerged as a compelling substitute technology for drug screening and personalized dosing. In this regard, Harada et al. established an oxaliplatin (L-OHP) resistant gastric cancer organoid (GCO) model and identified myoferrin (MYOF) as a potential gene by microarray analysis [137]. Recently, live imaging was used to analyze gastric cancer cell death, where homocellular spheroids were formed from the two-person (GC) cell lines, AGS and HGT-1. Heterocellular spheroids were created by co-culturing these cells alongside fibroblasts associated with cancer, building on the previous development of a combination of Doxorubicin and the cholesterol-reducing medication Lovastatin to explore the effectiveness of the drug on the model combination [138]. Drug-induced cytotoxicity was detected using a colorimetric assay with tetramethylazole salts and the growth and apoptosis of spheroids were also tracked using a real-time imaging and analysis system. The findings imply that the combination of Lovastatin and Doxorubicin is a successful approach in eradicating GC cells cultured either in 2D or 3D models. This further reinforces the potential of reusing Lovastatin as an adjuvant in Paclitaxel-based anticancer treatment (Figure 6A).

Figure 6.

3D GI disease modeling and drug screening. A i) Photographs captured in bright field using Incucyte live imaging of 6-day-old human gastric cancer cell line multicellular tumor spheroids (HGT-1 MCTS) at time intervals of 0 and 48 h after being treated with a concentration of 5 nM Docetaxel + 12.5 μM Lovastatin (D+L), captured at 0 and 48 h following the administration of 5 nM Docetaxel combined with 12.5 μM Lovastatin (D+L) treatment. ii) Continuous monitoring of heterocellular spheroids, comprising 250 HGT-1 cells and 250 cancer-associated fibroblasts (CAF), was conducted in real-time. 6-day-old multicellular tumor spheroids (MCTS) were subjected to single treatments of Docetaxel (5 nM), Lovastatin (12.5 μM), and a combined treatment of 5 nM Docetaxel combined with 12.5 μM Lovastatin (D+L). These treatments were then compared to the control group. iii) The impact of Docetaxel (D), Lovastatin (L), and the mixture of Docetaxel and Lovastatin (D+L) on bicellular MCTS viability was evaluated using the MTT assay after being treated for a duration of 48 h. Reproduced/adapted from Ref. [138]. B) A diagram outlining the procedure and parameters for differentiating human iPSCs to produce colonic organoids, along with a quantitative reverse transcription-polymerase chain reaction (qRT-PCR) dose-response curve of Imatinib, MPA, and QNHC on iPSC-derived lung organoids 24 h after SARS-CoV-2 infection. C) qRT–PCR based dose-response curve QNHC, MPA, and Imatinibon iPSC-LOs at 24 h following SARS-CoV-2 infection (reproduced/adapted from Ref. [139]).

Colon or colorectal cancer (CRC) is also a fatal malignancy in humans, and it is therefore essential to develop colon cancer models utilized in drug screening and clinical treatment. In this regard, Usui et al. successfully established an air-liquid interface (ALI) culture model using both healthy and cancerous human colon tissues, presenting a cystic arrangement incorporating the epithelial layer and adjacent mesenchymal stromal cells [140]. The findings revealed that the tumor model exhibited higher resistance to 5-fluorouracil and Irinotecan in comparison to the colorectal cancer cell lines SW480, SW620, and HCT116. This study may indicate that the ALI culture model derived from patients with colorectal cancer could be used as a tool for detecting chemotherapy resistance in the tumor microenvironment. Recently, iPSC-derived colon organoids (iPSC-COs) were developed for of inhibitors for SARS-CoV-2 using high-throughput screening of drugs was performed, and SARS-CoV-2 inhibitors were identified, including Quinacrine dihydrochloride, Mycophenolic acid, and Imatinib [139] (Figures 6B–C). Treatment with these drugs at the physiological level substantially suppressed infection of SARS-CoV-2 in iPSC-COs. SARS-CoV-2 infected iPSC-COs can be employed as a platform for investigating SARS-CoV-2 and its progression and screening of candidate drugs for the treatment of neo-crown pneumonia.

3.3. Kidney

The kidney is one of the primary organs of interest for the toxic effects of drugs, and the early stage of kidney damage is difficult to detect. Patients usually get to know when kidneys are severely damaged or even renal failure. According to statistics, 30% of candidates who participates in clinical trials during the development phase of new drugs must leave the clinical trial process due to unanticipated nephrotoxicity and adverse effects [141]. Therefore, a reliable nephrotoxicity prediction model can guarantee the safety of clinical drug use and significantly reduce the risk of drug development. The proximal renal tubule is a sensitive region of the kidney, and the metabolic excretion process of drugs usually happens on the epithelial cells of the renal tubule. Therefore, the early prediction of renal toxicity heavily relies on the proximal renal tubule.

3.3.1. 3D Models for Drug Screening for the Kidney

Human kidney epithelial cells (hKEpCs) obtained from nephrectomy samples demonstrated consistent capacity to generate kidney spheroids, when cultured in a suspended environment under non-adherent circumstances. Characterization of hKEpC spheroids exhibited increased expression of specific renal developmental markers in comparison to 2D [142]. For example, Kang et al. suspended immortalized mixed primary kidney cells in a mixture of ECM gel and Matrigel. Upon differentiation, these cells formed polarized tubules and exhibited protein uptake, enzymatic activity specific to kidneys, the response to hormones, and the operational capacity of drug transporters in the renal context, which makes it a valuable model for investigating renal physiology and nephrotoxicity [143]. Przepiorski et al. took iPSCs in suspension culture and used the Wnt agonist CHIR to form organoids and subsequently transferred the organoids to a rotating flask bioreactor for reaction, where hundreds of organoids containing renal tubular tissue grew and matured to resemble human fetal kidneys [144]. In another study, EBB was employed to create iPSC-derived kidney organoids, which demonstrated functional albumin uptake within the proximal tubules of the kidney structures [145].

The microfluidic-based kidney model also serves as an suitable platform for extended tissue culture, making it an ideal platform for drug testing purposes. The nephrotoxic drug, Cisplatin, and the nephroprotective drug, Cimetidine, showed predictable pharmacological effects, when kidney disease models on this microarray platform were used [146]. The glomerular chip developed by Zhou et al. could filter most inulin molecules, which indicated that the structure filtration function of the model closely resembled that of living human kidneys, demonstrating high and consistent performance [147]. Vormann et al. demonstrated a 3D perfusable proximal tubule microarray model consisting of renal proximal tubule epithelial cells. It was shown that the high activity of transporter proteins and the cobblestone morphology of cellular structures were important for the barrier function through inhibition studies of substrate uptake and transcellular transport [148]. In a different study, Petrosyan and colleagues employed a microfluidic chip in order to simulate the human renal filtration barrier by incorporating glomerular endothelial and human podocytes cells [149]. This promising platform offers great potential for understanding the mechanisms of glomerular diseases and performing drug screening investigations. A microfluidic chip consisting of three layers that provides a replicated setting for the kidney was also developed and assessed for drug-induced nephrotoxicity using peritubular capillary endothelial cells (PCECs) and renal proximal tubular epithelial cells (RPTECs), and three categories of drugs (cisplatin (DDP), gentamycin (GM), and cyclosporin A (CsA)) were subjected to experimentation on this model. It turns out that the microfluidic chip had better cell growth as well as drug nephrotoxicity evaluation compared to 2D [146].

Diekjürgen and Grainger have shown the early prediction of nephrotoxicity by an ex-vivo proximal tubule fragment culture. The 3D model exhibited higher levels of transport proteins and phenotypic status of epithelial cells at the time of renal metabolic excretion compared to 2D. The incidence of toxicity was found to be correlated by measuring the IC₅₀ values of 12 drugs compared to available pharmacological reports of clinical toxicity C_max values, which provides more reliable data for drug screening [150]. In a separate investigation, a 3D renal tubule model was exposed to Gentamicin and Cisplatin, the specific biomarkers of renal tubular injury, and it was observed that renal proximal tubular epithelial cells exhibited notably higher levels of kidney injury molecule-1 (Kim-1) and neutrophil gelatinase-associated lipocalin (NGAL) than usual [151]. Digby et al. used organoids and repeatedly administered them with low doses of Cisplatin for induction, with increased renal injury markers, inflammatory cytokine induction, and minor DNA damage [152].

3.3.2. 3D Disease Models for the Kidney

Kidney fibrosis is a prevalent pathological condition characterized by glomerulosclerosis and tubulointerstitial fibrosis, frequently leading to the progression of chronic kidney disease (CKD) or advanced-stage renal failure. For the pre-fibrotic glomerulosclerotic symptoms, conditionally immortalized human podocytes together with mature glomerular endothelial cells using a magnetic spheroid formation approach, as shown in Figure 7 [9]. Next, TGF-β and Adriamycin were added to create a fibrotic environment, and the subsequent addition of Nintedanib reduced podocyte loss. Its anti-fibrotic properties alleviated ECM dysregulation. Przepiorski et al. provided a new model for studying renal fibrosis, using iPSCs to generate organoids showing a pro-fibrotic environment after a prolonged culture, demonstrated by proliferation of interstitial cells and their transformation into pro-fibrotic myofibroblasts [144]. An FXR (bile acid receptor)/TGR5 (GPCR) dual agonist (INT-767) has been demonstrated to slow down the advancement of nephropathy and reverse renal injury, prevent renal fibrosis, and protect renal organs. Considering the poorly understood pathogenesis of renal fibrosis, renal organoids could provide an inexpensive, efficient, and useful tool for screening anti-fibrotic drugs [30].

Figure 7.

GlomSpheres were employed as a 3D spheroid model mimicking the kidney glomerulus, facilitating rapid drug screening. A) Imaging of the GlomSphere ultrastructure. i) In the SEM image of the complete GlomSphere, single cells were observable individually, along with tubule-shaped extensions marked by yellow arrows. ii) An SEM image reveals podocytes present on the surface of a GlomSphere. These podocytes feature short processes covering their surfaces, which overlap at the junction between cells (yellow arrow). iii) Podocytes (indicated by red arrows) were found surrounding the external surface of a tubule-shaped extension indicated by a marker in the form of a yellow arrow. The surface of the tubule seemed to be covered with foot-processes of podocytes. iv) In a TEM image, podocytes were observed on the GlomSphere surface, showing interdigitating structures. v) The TEM image illustrates interdigitating podocytes, along with accumulated fibrillar collagen within a lumen-like space (indicated by a yellow arrow). vi) Podocytes displaying interlocking formations and fibrillar, unstriated collagen found within a cavity-like area. vii) A fibrillar collagen Type IV layer (indicated by a arrows in yellow color) was placed within the extracellular region situated between an adjacent podocyte and glomerular endothelial cells (GEnCs). B) Inducing and regulating fibrotic phenotypes in GlomSpheres: i) The image shows an untreated GlomSphere through a projection with maximum intensity with podocytes (depicted in green) were observed surrounding a central core of glomerular endothelial cells (GEnCs) in a peripheral manner. Collagen Type IV was stained (appearing in blue) and predominantly found at the interface between podocytes and glomerular endothelial cells (GEnCs). ii) After being incubated for 72 h with Adriamycin (10 nM) and TGF-β1 (10 ng/ml), the GlomSphere exhibited a noticeable decrease in podocyte coverage on its surface. Additionally, the fluorescence of Collagen Type IV was potentially reduced. iii) After 72 h of incubation with Adriamycin (10 nM) and TGF-β1 (10 ng/ml), the GlomSphere exhibited a noticeable decrease in podocyte coverage on its surface. Additionally, the fluorescence of collagen Type IV was potentially reduced. iv) A GlomSphere without any treatment. V) After being incubated for 72 h with TGF-β1 (10 ng/ml) alone, the GlomSphere exhibited an apparent increase in collagen Type IV (appearing in blue) expression, and it was not predominantly localized at the interface between podocytes and GEnCs. However, there was no noticeable change in podocyte (appearing in green) coverage. vi) After 72 h of incubation with Nintedanib (10 μM) and TGF-β1 (10 ng/ml), the GlomSphere showed a decrease in collagen Type IV (appearing in blue) fluorescence compared to the TGF-β1-only condition (reproduced/adapted from Ref. [9]).

Polycystic kidney disease is a prevalent genetic disorder that is often passed down through generations, mainly divided into autosomal dominant polycystic kidney disease (ADPKD) and autosomal recessive polycystic kidney disease (ARPKD), with ADPKD being the most prevalent variant [153]. ADPKD is identified by the presence of genetic mutations in the PKD1 and PKD2 genetic sequences. Kidney cysts gradually form and enlarge during the course of the disease, eventually leading to kidney failure [154]. The prevalent approach to investigate this disease is through the utilization of kidney organoids as models and allow screening for drugs with unknown effects on cyst formation. In this regard, researchers used Bblebistatin, a non-muscle myosin II inhibitor that can strengthen and tighten tubules and prevent them from becoming cysts to induce encapsulation in PKD organoids [155]. It emphasized the ability to clarify essential mechanisms of PKD and made it possible to explore new PKD pathways and screen for new drugs. Wu et al. performed 3D cyst experiments and identified resveratrol as beneficial for the PKD treatment and NF-κb as a potential therapeutic target [156]. Osafune found that inhibitors of the target of rapamycin (mTOR) proteins prevented renal cyst formation in organoid models [157]. In a different research effort, Booij et al. established a 3D cyst culture model for the screening of drugs and then determined complex therapeutics by phenotypic analysis to detect inhibition of PKD targets of action, such as CDK, CHK, and AuroraA. These molecules have the potential to prevent forskolin-induced cyst swelling with the help of limited proliferation [158]. Generally, most of the current studies on polycystic kidney disease have been conducted using organoid models to establish drug screening methods for the discovery of therapeutic agents and potential targets.

Recently, COVID-19-associated acute kidney injury (COVID-AKI) was observed to be a common complication in individuals with SARS-CoV-2 infection. In this regard, Rahmani et al. used iPSC-derived kidney organoids and found that the angiotensin II receptor blocker Cloxacin may play a role in protecting susceptible proximal tubular cells during SARS-CoV-2 infection [159]. Working in the same direction, Lon et al. demonstrated the effectiveness of selective HDAC8 inhibitors in AKI models, particularly the therapeutic potential of this target for AKI through biochemical analysis, zebrafish AKI phenotyping, and analysis of human kidney organoids [160].

3.4. Heart

It is well known that drug-induced cardiotoxicity can result in severe adverse cardiovascular events, like myocardial necrosis, myocardial infarction, and severe fatal arrhythmias, which leads to the termination of drug development or its removal from the market [161, 162]. Terfenadine, an antihistamine, is an example of such kind of drug, which obstructs cardiac potassium channels, leading to the extension of the action potential and resulting in a significant lengthening of the QT interval observed on the electrocardiogram (ECG). Ultimately, this leads to severe arrhythmias and can result in death [163, 164]. Antibiotics, tranquilizers, and antidepressants are some other medications with comparable actions [165, 166]. Cardiotoxicity is often evaluated in cell cultures and does not provide a true picture of human tissue. On the other hand, animal models lack translational relevance to humans, so employing 3D human cardiac models is a preferred approach.

3.4.1. 3D Models for Drug Screening for the Heart

It is well known that human primary cardiomyocytes are terminally differentiated, so these cells cannot expand in culture [167]. Acquiring a sufficient number of cardiomyocytes would require obtaining heart tissue biopsies from patients, which is unacceptable. Other cell sources can be derived from hESC or iPSCs. Until now, most of the cells used in cardiovascular tissue engineering are from iPSCs [168].

Currently, it is desirable to assess drug response in vitro using fabricated cardiac models. Generally, iPSC-derived cardiomyocytes (iPSC-CMs) form spheroids by intercellular adhesion on a low-adhesive surface. Spheroids can secrete laminin, ECM fibronectin, and collagen Type I and IV and express cardiac-specific markers (MLC2v, MEF2C, and cTnT), suggesting maintenance of heart tissue-specific phenotype and function [169–171]. In a study, cardiac spheroids (200–400 μm in size) were prepared from iPSC-CMs for evaluation of drug response, where Isoprenaline increased the beating rate of spheroids while Blebbistatin (a cell permeability inhibitor acting on non-myosin II ATPase) decreased the amplitude of spheroid motion. However, the spheroid structures were unstable as spheroids were small and fragile and could not withstand prolonged measurements [172].

In addition to spheroid models, scaffold-based models, such as iPSC-CMs loaded into collagen and fibrin, are also common [173, 174]. By seeding cells into a scaffold, iPSC-CMs adhere, survive, proliferate, and develop functional characteristics. For example, a heart model with an excitation-contraction function was successfully constructed by embedding fibroblasts and iPSC-CMs within a hydrogel and culturing them for a few days [174]. Isoprenaline was also found to increase myocardial contractility and the addition of E-4031 (a potassium channel blocker) caused spontaneous irregular beats consistent with arrhythmias. These results demonstrate that this model can accurately assess the impacts of the drug. The arrangement of iPSC-CMs was also controlled by seeding cells on neatly aligned polylactide-coated nanofibers [175]. The study demonstrated that 3D cardiac tissue-like constructs accurately replicated both inherent structural and functional attributes of the cardiac tissue.

There are also advancements in cardiac models fabricated using bioprinting. For example, Nakayama and coworkers, utilized a 3D bioprinting technique that can fabricate spheroid-based structures on microneedle arrays, where spheroids were strung onto microneedles and laminated together to fabricate tissue patches [176]. In another study, 3D cardiac micro-tissues were generated using an in-house developed micro-continuous optical printing system, which enables the creation of these micro-scale tissues in seconds with precise micro-scale alignment and the ability to fit them into a 96-well plate. These micro-tissues displayed strong sarcomere organization and a significant increase in maturity marker expression after seven days. The cardiac micro-tissues were then subjected to validation with two representative drugs, isoproterenol and verapamil, at various doses, revealing corresponding and measurable changes in beating frequency and displacement. This approach presents a promising solution for high-throughput drug screening in cardiac research [177]. Similarly, Arai et al. 3D bioprinted a tubular heart model that form rhythmic contractions, which was used for drug testing including Isoprenaline, Propranolol, Blebbistatin, and Doxorubicin [178].

As mentioned above, the successful construction of 3D human cardiac models for preliminary drug testing and prediction of cardiotoxicity is a feasible and cost-saving option. For example, Lu et al. used 3D cardiac tissues constructed from iPSC-CM in silicone molds and successfully validated its ability to allow predictive assessment of cardiotoxicity at the tissue level using rapid kinetic fluorescence imaging and quantitative analysis of contraction curves [179]. Several classes of drugs, including the antidiabetic drugs Metformin, Rosiglitazone, antibiotics Ampicillin, Erythromycin, Tevafltrovafloxacin and the anticancer drugs, such as Vandetanib and Tamoxifen, were tested, where the results were consistent with clinically-observed cardiotoxicity. In the same line, Weng et al. constructed a microfluidic chip to co-culture iPSC-derived endothelial cells (iPSC-ECs) and iPSC-CMs alongside tumor spheroids simultaneously [180]. The cardiotoxicity of two anticancer drugs, Oxaliplatin and Adriamycin, was analyzed by examining their influence on the inherent rhythm rate and propagation speed of cardiac tissues activity derived from iPSCs. The results showed consistency with in-vivo observations.

3.4.2. 3D Disease Models for the Heart

More than 12 million patients worldwide are at a risk of suffering from cardiac arrhythmias, and the associated management costs exceed $30 billion placing a significant burden on the healthcare systems [181]. However, there have been no new advances in arrhythmia treatment, partly due to differences between species in disease models, including heart rate, myofilament composition, and cardiomyocyte electrophysiology [182] in addition to the poor understanding of mechanisms of arrhythmia initiation, maintenance, and termination. More promisingly, iPSC-CMs have driven the progress related to arrhythmias in 2D and 3D, especially the 3D arrhythmia model that can function as a platform for screening antiarrhythmic drugs [183, 184]. Williams et al. conducted a study primarily focused on on advances in microarray technologies and on the principle that β-cyclodextrin (MBCD) can break the cell membrane and block calcium channels, causing arrhythmias, resulting in the development of steady 3D representations of cardiac arrhythmias [185]. In this study, human cardiac fibroblasts and iPSC-CMs were seeded in fibrin at a ratio of 3:1, which mainly considered relative proportion of cardiomyocytes (CMs) to non-cardiomyocytes in the human heart [185, 186]. After seven days of culture, compact engineered heart tissues were formed, arrhythmias were induced with a high concentration of MBCD, and real-time functional measurements were performed on a previously described cardiac microchip. Complex electrophysiological activity began to appear on Day 5, with multiple wave peaks and unsynchronized contractions resembling fibrillation identified in the mapping. Treatment of this arrhythmia tissue with Lidocaine classified as a a class Ib antiarrhythmic drug, Isoprenaline, Flecainide, categorized as a class I antiarrhythmic agent, Propranolol, classified as a class II antiarrhythmic agent, and Amiodarone, designated as a class III antiarrhythmic agent produced significant electrophysiological changes related to the mechanisms (Figure 8). This study brings new advances to in-vitro arrhythmia models, simulating the intricate electrophysiological intricacies of tissues experiencing arrhythmias, which effectively reproduced the action characteristics of various drugs and are anticipated to function as a research tool for investigating the pathophysiology of arrhythmias and pharmacology in the future.

Figure 8.

A) A diagram illustrating the process of generating and treating tissue using cardiac fibroblasts and iPSC-CMs obtained from normal individuals. These cells were mixed in fibrin and placed between two flexible polymeric rods. B) At Day 5 after the treatment, the bottom section of electrophysiological phase maps represents the arrhythmic condition, while the top section depicts the time-matched control. White arrows indicate the wavefront direction. The key located on the right-hand side explains the colors representing different stages of the calcium transient within a ventricular cardiomyocyte (scale bar, 500 μm). C) The phase maps of an irregular tissue are presented, illustrating the condition before (on the left) and after 30 min of Lidocaine incubation (on the right). White arrows indicate the wavefront direction. The key located on the left side of the Figure 8C provides a description of the color-coded depiction of specific action potential phases, accompanied by a scale indicator measuring 500 μm. D) The number of wavefronts on average per frame, the speed of conduction (measured in centimeters per second), and the duration of the cardiac cycle in irregular heart tissues before and after being subjected to anti-arrhythmic drugs, namely Lidocaine and Isoproterenol, were compared (reproduced/adapted with permission from Ref.[185]).

Another study was conducted on the Torsade de Pointes (TdP) model of torsional ventricular tachycardia [187], a polymorphic ventricular tachycardia (TdP) that can lead to cardiac arrest and death. It is characterized by distorted waves in ECG and leads to ventricular fibrillation, eventually causing death, which is mainly caused by drugs. The authors fabricated a 3D iPSC-derived cardiac tissue to model TdP arrhythmias. The constructs were fabricated as floating sheets containing a mixture hiPSC-CMs and mesenchymal cells, which were verified by testing the levels of ion channel gene expressions that were like those of the adult heart except for KCNJ2 and SCN5A. Similarly, its waveforms and testing its response to E-4031 resulted in the observation of a sustained tachycardia waveform and a characteristic TdP waveform. This study has positive implications both in terms of visualizing the occurrence of TdP-like arrhythmias and drug screening and testing for treating arrhythmias.

Atherosclerosis is the primary underlying factor responsible for coronary heart and cerebrovascular diseases [188]. It ranks among the top causes of morbidity and mortality on a global scale and is primarily marked by the buildup of fatty deposits within the subendothelial space of blood vessels [189]. Atherosclerotic plaques originate and advance due to the accumulation of macrophages (foam cells) laden with low-density lipoproteins, accompanied by changes in ECM organization. These factors contribute to the narrowing of the local vessel, known as stenosis. Mimicking atherosclerotic conditions were confirmed by the ability to uptake oxLDL (oxidized low-density lipoprotein) and the expression of relevant receptors, along with their secreted cytokines, laying a foundation for advanced models for developing future anti-atherosclerotic drugs.

Pulmonary arterial hypertension (PAH) arises primarily due to the abnormal and excessive growth of smooth muscle cells in the pulmonary arteries, specifically termed as pulmonary artery smooth muscle cells (PASMCs). This excessive growth leads to the thickening of the inner layer of blood vessels, and the histopathological marker of this pulmonary vascular remodeling is medial thickening. Furthermore, the restructuring of pulmonary blood vessels leads to the narrowing of smaller pulmonary arteries, as a consequence, this leads to increased resistance in the pulmonary blood vessels, causing an escalation in pulmonary artery pressure. Eventually, this results in right-sided heart failure [190]. Morii and Tanaka et al. seeded PASMCs from PHA patients in a gelatin/fibronectin hydrogel to obtain a 3D-PAH model [191]. Then, the addition of platelet-derived growth factor-BB (PDGF-BB) was performed to induce thickening in 3D-PAH. This step was undertaken based on increasing evidence suggesting that the initiation of PDGF signaling has a substantial impact in the progression and enhancement of pulmonary arterial hypertension. When treated with Imatinib, a PDGF signaling inhibitor, along with the endothelin receptor antagonist Bosentan and the phosphodiesterase-5 inhibitor Tadalafil, the thickness of 3D PAH tissue was reduced, accompanied by other improved properties [192]. These results suggested that the 3D-PAH model treated with PDGF-BB can be used to screen anti-PAH drugs to some extent, bringing great hope for new drug development.

3.5. Bone and Cartilage

Numerous studies have shown that cells in 3D models have a different morphology than cells in 2D. Additionally, it is widely known that cell morphology significantly influences various including cell growth, maturation, programmed cell death, and the synthesis of genetic material and proteins [124, 193, 194]. Compared to 2D cultured osteoblasts and chondrocytes, 3D culture has a lower cell proliferation rate [195]. Moreover, multiple studies have demonstrated that 3D models have the capability to enhance osteogenic or chondrogenic effects [196–198], and cells in 3D usually demonstrate a gene expression profile largely resembling that for in-vivo tissues [199–201]. Different bone and cartilage models were also found to have different drug sensitivity. Cells typically exhibit greater resistance to treatment in 3D than 2D models, which may be a better vehicle for in-vivo drug response [202, 203].

3.5.1. 3D models for Drug Screening for Bone and cartilage

In the field of bone and cartilage tissue engineering, spheroids are the gold standard. They have been shown to promote bone and cartilage differentiation in vitro and also hold potential for drug screening [204]. For example, an in-vitro study evaluated the effects of zoledronic acid (ZA) on osteoblast (hFOB 1.19) spheroids [205]. The findings revealed that ZA had dual dose-dependent impacts on osteoblast functions in the 3D model. At higher micromolar concentrations, it led to growth arrest and apoptosis of MSCs. However, at lower concentrations, it promoted cell proliferation to a certain extent. By restoring cellular interactions with neighboring cells and the ECM, the spheroid model mimicked the native tissues more accurately than the 2D model did, making spheroids-based assays to increase their predictive ability for drug responses. Another work presented a simple approach for fabricating a high-throughput concave microwell array, which allowed the formation of heterocellular spheroids comprising chondrocytes [206]. This method also enabled the maintenance of the characteristic traits and functional abilities of articular chondrocytes within the spheroids. Compared to the conventional 2D approach, 3D spheroid culture was noted to enhance and maintain the characteristics specific to cartilage and functional capacity. The developed method offers a valuable platform with diverse applications in cartilage biology and high-throughput drug testing.

Mechanical properties of hydrogels are a key factor in cytoskeleton formation. Several osteoblastic cell lines (CAL72, SAOS2, and MG-63) along with human primary osteoblasts cultured within a hydrogel composed of silylated hydroxypropyl methylcellulose showed higher viability and proliferation, and better osteogenic differentiation than cells cultured in monolayer [207]. Although numerous hydrogels have demonstrated promising outcomes in human osteoblast cultures, experiments with materials have demonstrated that osteogenic culture with hydrogels is still in their early stages of research [208, 209].

Mature cartilage is a highly organized viscoelastic substance, characterized by pronounced acellular properties in comparison to many other tissues [210, 211]. Due to this, research studies have endeavored to develop alternative 3D cartilage models using hydrogels, aiming to mimic essential characteristics present in native cartilage. The primary benefit of hydrogels is that they allow for precise control over cell density, which is especially important for models with a low cell count (e.g., articular cartilage). Simple hydrogel systems are highly convenient for drug screening, and it has been demonstrated that cell supporting platforms made by hydrogels can be rapidly created [203, 212, 213]. More recently, the advantages of combining spheroids and hydrogels have been integrated via encapsulation of spheroids in hydrogels to create a hybrid model [214]. This model can be easily adapted for chondrocyte culture and drug screening.

Working in this direction, researchers utilized a blend of alginate and gelatin methacrylate (GelMA) to prepare microfluidic-based systems, and loaded osteoblasts showing promising viability and high levels of protein expression, such as bone morphogenetic proteins and Type I collagen [215]. Furthermore, An et al. developed a technique based on microfluidics to sequentially encapsulate MSCs using alginate microgels at a single cellular level demonstrated considerably improved osteogenic effects, leading to accelerated in-vitro mineralization and improved capacity for bone formation in vivo [216]. Current microfluidic chips are also widely used for in-vitro drug detection of skeletal system diseases [217–219]. The potential of organ-on-a-chip technology shows potential as a viable substitute for conventional 3D models of cartilage. For example, a microfluidic system was recently developed by Rosser et al. with the ability to generate periodic shearing and variations in concentration, replicating the conditions that closely resemble the avascular tissue on and beneath the articular surface [220]. The cells within this system maintain their circular morphology and express cartilage-related genes. Moreover, the incorporation of pro-inflammatory cytokines into the system showcases its capability as a screening platform [221]. This model more closely encapsulates the endogenous joint environment and highlights its capacity a framework for evaluating drugs.

Bioprinting technology, on the other hand, has tremendous advantages in producing substitutes that mimic healthy hyaline cartilage and bone [222]. For example, EBB was employed to construct a 3D bone model to investigate the effects of four drugs [223]. The results showed that the activity of alkaline phosphatase (ALP) and the expression of genes related to osteogenesis exhibited higher levels than those in 2D model suggesting that the bioprinted cartilage holds potential for drug testing.

3.5.2. 3D Disease Models for Bone and Cartilage

Disease models play a crucial role as essential tools to investigate the molecular mechanisms responsible for disease development and contributing to the progress of new treatments for bone-and cartilage-related conditions [224].

Multiple myeloma is among the most widespread hematologic cancers globally. This form of cancer falls under the classification of a plasma cell carcinoma, distinguished by anemia, elevated calcium levels, invasive malignant bone infiltration, and renal dysfunction [225]. The imbalance of normal bone homeostasis caused by multiple myeloma can result in the development of myeloma bone disease [226]. Although the exact mechanism through which plasmacytoma disrupts bone remodeling is not fully understood, conventional bone homeostasis processes have been extensively investigated [227–229]. Previous studies have developed 3D models that mimic multiple myeloma and employed these to examine how cell-cell interactions impact resistance to drugs in multiple myeloma. For example, Belloni and colleagues utilized a 3D rotating bioreactor for co-culturing myeloma cells alongside bone marrow cells within a gelatin scaffold. They showed that the reconstructed multiple human myeloma microenvironment might represent an effective platform for conducting drug tests and exploring interactions between tumors and the surrounding stromal cells [230]. Recently, Monteiro et al. presented a 3D osteosarcoma (OS) model, as shown in Figure 9, in which they generated the tumor structure encapsulating OS spheroids in hydrogels, co-cultured with mesenchymal stem cells and osteoblasts and evaluated the capabilities of the osteosarcoma model for drug development by Adriamycin-responsive dose testing, showing that this model had good predictivity in terms of drug resistance [231]. In another study, Voissiere et al. developed human chondrosarcoma HEMC-SS spheroids, which allowed to achieve valuable knowledge about the biology of tumor cells and their response to drugs [232]. These spheroids produced chondrogenic ECM, which was abundant in glycosaminoglycans and collagen Type-II. The reaction of chondrosarcoma spheroids to the hypoxia-activated prodrug TH-302 and Doxorubicin was contrasted with their response in a 2D setting. As anticipated, they showed greater cytotoxic potency when targeting larger hypoxic spheroids on the 14^th day, in comparison to non-hypoxic spheroids (Day 7). Larger spheroids demonstrated increased resistance to Doxorubicin. These larger spheroids offer an improved in-vitro chondrosarcoma model to assess the impact of drugs aimed at the tumor.

Figure 9.

A) An illustration depicting human methacryloyl platelet lysates (PLMA) hydrogel production and the creation of an in-vitro OS model utilizing a scaffold-based approach. B) Visualization of the arrangement and orientation of tumor and stromal cells in a co-culture model, where confocal images were acquired for both the control and Doxorubicin (DOX) treated (at 1.1 μM) co-culture model, employing cells that were altered genetically with fluorescent protein (red fluorescent protein labeled MG-63, green fluorescent protein labeled human bone-marrow mesenchymal stem cells (hBM-MSCs)) demonstrating the synergistic interaction between MG-63 tumor cells and stem cells over a 24-day culture duration. C) LIVE/DEAD images of the 3D OS model, portraying the survival of MG-63 tumor spheroids grown with and without hBM-MSCs and FhOBs at both the 14-day and 24-day culture time points. Cell responsiveness to a 3-day treatment with DOX (1.1 μM) was evaluated on the 24^th day (adapted with permission from Ref. [231]).

Osteoarthritis (OA) stands as the prevalent joint condition, significantly impacting the well-being of the aging population worldwide. Still, no better preventive measures or treatment options are available [233, 234]. Patients with OA exhibit age-related characteristics in their articular cartilage, and chondrocytes release senescence-related secreted factors like pro-inflammatory cytokines and enzymes that degrade the ECM. These factors hold considerable importance as being significant mediators in promoting the progression of OA [235]. In a recent research, Yang and colleagues investigated the therapeutic efficacy of ABT263 in post-traumatic OA by drug-selective removal of senescent cells, which they further corroborated by testing the drug ABT263’s capability to remove senescent cells in vitro using monolayer and spheroid cultures of osteoarthritic chondrocytes [236]. Recently, an OA disease model was reported based on goat cartilage explants. Three different drugs (i.e., BMP7, Rapamycin and Celecoxib) were used to validate the developed drug, all of these studies showed that disease improvement in the explant model was dependent on the drug concentration, leading to a reduction in discrepancies observed between preclinical and clinical studies in the process of developing drugs for OA [237].

Similarly, osteoporosis is also a common systemic bone disease that can lead to bone fragility and increased fracture risk, with a high prevalence in the elderly and menopausal women [238]. It is a degenerative pathology for which no good measures exist. Even with pharmacological treatment, there is a problem of long-term dependence; therefore, many studies were recently published to find appropriate options to treat osteoporosis [239]. For instance, Breathwaite and colleagues developed an in-vitro bone tissue model to assess four drugs’ effects on osteogenesis. Icariin and Purmorphamine were expected to promote osteogenesis, while PD98059 and U0126 were predicted to hinder it. They cultured BMSCs that were differentiated in scaffold-free bioprinted structures and in 2D culture for the purpose of comparison [223]. For each construct, spheroids were bioprinted in layers following cuboidal configuration, which then effectively fused and the bioprinted constructs were taken off from the needle array and placed onto low-adhesion plates. The findings indicate that the bioprinted model provides a drug screening platform that is more dynamic and aligned with biological processes compared to the conventional 2D monolayer culture.

Rheumatoid arthritis (RA) is a systemic autoimmune condition that is a common mode of presentation in inflammatory arthritis. It is a proliferating synovial tissue dominated by macrophages and fibroblast-like synoviocytes. While significant research has been conducted on the pathophysiology of RA, a comprehensive understanding of its exact cause is still lacking. Consequently, current therapies for RA are limited to alleviating the clinical symptoms or slowing down the advancement of the disease [240, 241]. To gain a deeper comprehension of the mechanism behind cartilage damage in rheumatoid conditions, Peck et al. developed an in-vitro model comprising synovial fibroblasts, macrophages, and primary chondrocytes mixed together with gelatin microspheres in an alginate solution, which was shown by immunostaining and biochemical analysis to key pathological characteristics found in RA [242]. The administration of Celecoxib as a drug treatment illustrated that the model was effectively responded to the drug by reversing the cartilage damage.

3.6. Brain

Approximately one in six individuals globally are affected by neurological disorders, encompassing conditions, such as schizophrenia, epilepsy, Alzheimer’s disease, brain tumor etc [243]. Currently, researchers mainly depend on the conventional animal models with several limitations, like being expensive and time-consuming, and the treatments in animals often fail in clinical trials so understanding of the cause and progression of neurological disorders drives researcher to seek for alternative models to better shed light on neurological disorders and discover potential treatments or drugs. To simulate the complexity of 3D neural tissues, neural organoids are the main models in the context of drug discovery.

3.6.1. 3D Models for Drug Screening for Brain

Recently, different 3D neural organoids, including spinal cord, midbrain, hippocampal, neocortex, and cerebral organoids, have been constructed and showed their potential for neurotoxicity testing. For example, Lee et al. used three consecutive steps to establish a spinal cord organoid and then analyzed the effects of six antiepileptic drugs (Phenytoin, Carbamazepine, Gabapentin, Primidone, Lamotrigine, and valproic acid) on neural tube defects (Figure 10) [244]. Their observations indicated a delay in the start of morphogenesis, and there was a failure in the closure of the neural tube when treated with valproic acid or Carbamazepine. At the same time, the treatments by other drugs maintained normal morphogenesis. These findings closely aligned with previous clinical data regarding the risks associated with antiepileptic drugs.

Figure 10.

A) Illustrations depicting the process of generating human spinal cord organoids (hSCOs). B) Images were taken in bright-field of hSCOs treated with drugs (reproduced/adapted with permission from Ref. [244]).

Another automated midbrain organoid model was used to conduct a screening of a library comprising 84 compounds, which was successfully recognized known nigrostriatal toxicants, and identified 3,3’,5,5’- Tetrabromobisphenol A as the toxicant to target dopaminergic neurons [245]. Along the same line, Monzel et al. created a neurotoxin-induced Parkinson organoid model and confirmed its toxicity prediction by treating with 6-hydroxydopamine. Thus, they demonstrated the valuable midbrain organoid as a Parkinson’s disease model in vitro and its potential use as a platform to test putative neurotoxic compounds [246]. Recently, iPSCs were used to generate choroid plexus organoids to investigate the capacity of SARS-CoV-2 to invade brain cells, and the findings indicated that the epithelial cells of the choroid plexus were robustly infected, while the neurons and astrocytes were hardly infected. Furthermore, the authors found that combined with the infection, there was also increased cell death and cellular dysfunction [247].

3.6.2. 3D Disease Models for Brain

Collectively, iPSCs-derived brain organoids could be used as a platform to further investigate the mechanism of brain dysfunction and develop or screen relevant treatment strategies. Neural organoids are expected to develop as the next generation of high-throughput screening tools, offering more physiologically-relevant predictions for neurotoxicity assessments [248]. However, distinguishing and quantifying the relevant data still remains a challenging task. Recently, Henrik et al. presented a method to produce highly homogeneous midbrain organoids with morphology, size, cellular composition, and structure. The resulting organoids have the key features of the human midbrain. By automating the entire workflow, drug effects could be assessed at the single-cell level [249].

Glioblastoma is the most common and deadliest brain cancer in adults. Researchers study glioblastoma, a fast-spreading brain tumor, and its cancer stem cells that resemble embryonic stem cells in terms of growth [250]. These cells exhibit heterogeneity and are challenging to characterize. Researchers use advanced tools, such as organoids and bioprinting, to understand cancer stem cells [251, 252]. In a recent study, Tang et al. used 3D bioprinting to create glioblastoma models, which is the most common and deadliest brain cancer in adults [253, 254]. In glioblastoma, a prominent contribution to the tumor mass is made by macrophages/microglia. In this study, the function of macrophages in 3D was assessed by comparing the growth of glioblastoma stem cells (GSCs) alone or with astrocytes and neural precursor cells in a hyaluronic acid-rich hydrogel, with or without the presence of macrophages. The inclusion of macrophages in bioprinted constructs reproduces patient-specific transcriptional profiles. Unique molecular dependencies in GSCs, relative to cell-laden gel culture, were identified through whole-genome CRISPR screening conducted with 3D bioprinted constructs. This indicates that bioprinted models can serve as a scalable and physiologically-relevant platform for studying brain tumor conditions.

4. Future Consideration for Establishing a Physiologically-relevant Microenvironment

The tissue architecture and mechanical properties have a vital role in regulating a diverse range of biological consequences, encompassing stem-cell differentiation, cancer progression, and the response to anticancer treatments. Microenvironmental cues, including ECM components, stiffness, etc., govern cell behavior and determine the cellular response to the surrounding ECM environment [255]. To create a bionic 3D model for effective drug screening, the appropriate combination of cells, biomaterials, fabrication modalities, and stimulators are essential. Generally, the microenvironment can be reflected by physical, chemical, and biological features of tissue. The physical features usually include ECM alignment, stiffness, and tissue microarchitecture. Chemical features include ECM components and cytokines, and biological features include cellular composition and intercellular actions, autocrine and paracrine, vasculature and innervation, etc.[101]. To design an effective an in-vitro setup for screening drugs, these three characteristics should be deliberated for the selection of cells, biomaterials, fabrication modalities, and subsequent tissue maturation process.

4.1. Cells Sources

Typically, three types of cell sources are commonly utilized in constructing 3D tissue models encompassing primary cells, cell lines, and cells originating from stem cells. Primary cells are acquired directly from human tissues and serve as excellent candidates for replicating specific tissue functions. Nonetheless, primary human cells are in short supply, and their ability for further subculture is restricted. Cell lines can be passaged repeatedly in vitro, and they are cheaper and easily accessible, so cell lines are suitable alternatives when utilized for investigating simple or specific characteristics. Stem cell-derived cells are widely accepted in construction for 3D tissue models. These include MSCs, ESCs, and iPSC-derived cells. iPSCs are becoming increasingly popular because of their potential capable of differentiating into nearly any cell type within the body given appropriate stimuli and showing donor to donor variation. Nevertheless, the applications of these cells are often limited by their incapability of functional maturation enough as adult cells [256]. In recent times, co-culture platforms have been garnering more attention because they offer improved support for tissue mimicry and function. Non-parenchymal cells, including fibroblasts and endothelial cells, are commonly used to support parenchymal cells or exert paracrine effects [256–258]. Immune cells are also integrated into tumor [259] or inflammation models [260, 261].

It is widely recognized that many drugs only exhibit effectiveness in a specific subset of patients with certain diseases, while causing various adverse effects among these individuals. Patient-specific genomic conditions that impact varied drug effectiveness and potential side effects encompass detrimental mutations and diverse expression levels of drug targets and proteins involved in drug metabolism [262]. Personalized medicine calls for combining personalized drug screening (PDS) with detailed personal health information to expedite the identification of effective therapeutic compounds that offer greater efficacy and reduced toxicity [263]. These models can be created using human cells affected by diseases or dysfunctions for examining disease mechanisms and assessing the effectiveness of novel compounds. In the literature, diverse models in both 2D and 3D are created for personalized drug screening and disease modeling, with a primary focus on cell sources specific to individual patients, which encompass cells derived from iPSCs or primary cells from individuals with the disease [264, 265]. Currently, the availability of primary cells obtained directly from patients is restricted, and their utilization is not widespread. Recent studies have demonstrated the use of cells derived from induced iPSCs in creating various tissue models [266]. In particular, iPSC-derived organoids have been used for genetic disease models and PDS [267–269]. However, achieving the functional maturity of iPSC-derived cells comparable to adult cells remains a significant challenge. There is a significant demand for research focused on enhancing iPSC differentiation and maturation [270, 271]. In the future, utilizing primary diseased cells from patients or iPSC-derived cells to construct co-culture platforms will offer further insights into personalized disease modeling and PDS development. Moreover, integrating the findings from PDS and personal genome analysis will significantly advance the progress of precision medicine [272].

4.2. Biomaterials

Biomaterial selection is another important consideration for designing tissues with specific physical and chemical characteristics that are desired. Naturally-derived biomaterials are appealing because of their similarity to the native ECM in regards of biophysical and biochemical characteristics. These substances hold the capability to support cell attachment, growth, specialization, and movement. Nevertheless, natural biomaterials often possess limited mechanical strength. In contrast, synthetic materials offer high precision, easy modification, and reproducibility, allowing for control over their biological and mechanical characteristics. However, replicating all components of the native ECM still remains challenging [273, 274]. To overcome these challenges, researchers have utilized composite hydrogels that combine natural and synthetic biomaterials. Typically, synthetic materials are incorporated to enhance mechanical resilience, while naturally sourced materials are incorporated into the hydrogel structure to enhance the biological activities of cells [275]. Moreover, many naturally-derived materials like HA, collagen, and gelatin only represent individual elements of the ECM and do not include additional essential components like glycosaminoglycans, elastin, proteoglycans, and laminin. Recently, there has been increasing interest in the field of tissue engineering regarding the application of dECM obtained from tissues and organs [276–278]. Research has shown that dECM exhibits unique compositions, and cells react to these matrices in a manner specific to the tissue, which has a vital function in preserving specific phenotype and functionality. These results illustrate the adaptability of dECM in constructing tissue that closely replicate the in-vivo microenvironment of tissues and organs in an appealing manner [277, 279].

Nevertheless, dECM is usually derived from animals, which may bring risks of viral transmissions or other diseases [280, 281]. The selection of biomaterials for 3D models is contingent on several factors, including the biofabrication technologies, tissue of interest, drug mechanisms, etc. Future research is essential to develop highly tunable materials capable of delivering the desired biological, mechanical, and chemical characteristics to faithfully mimic the tissue microenvironment.

4.3. Biofabrication technologies

Scaffold-free models, like spheroids, are widely employed as the most common 3D models for drug screening applications. This popularity is attributed to their simplicity and ease of integration into existing 2D workflows [282]. The adaptability of these models to 384-well and 1536-well plate formats allows for comprehensive testing and analysis utilizing conventional laboratory equipment and existing assay readouts [283, 284]. However, the lack of exogenous ECM in scaffold-free models sometimes constrains their suitability in relevant drug testing [285]. Simple matrix-embedded 3D models demonstrate spheroid or organoid generation and can be readily adjusted to fit well-plate layouts and liquid handling automation [286, 287]. Co-culture models can also be created by embedding mixed cell types in hydrogels, presenting a better tissue microenvironment with heterogeneity and complexity as in vivo [288]. These matrix-embedded modes were limited by the incapability of combining mechanical stimulations, such as flow shearing, which can be addressed by utilizing microfluidic chip models. Microfluidic-based have been extensively employed for drug testing purposes due to their advantages, such as providing mechanical stimulations by flowing a subtle dose of tested drugs [289, 290]. Besides, multiple organs can be assembled into a human-on-a-chip platform, reflecting the interaction of different organ-processing drugs [291]. To produce sophisticated 3D models and facilitate interactions between cells and ECM, as well as between cells themselves with tuned matrix and delicate structures, manual approaches are mostly labor-intensive, through which the created models are hardly suited to well-plate formats for high-throughput screening purposes. Thus, those models are often limited to applications in disease modeling for foundational biological investigations [4].

On the other hand, 3D bioprinting has rapidly gained prominence as a highly promising method for effectively producing 3D models in a microplate format. It permits the integration of tunable ECM elements to attain in-vivo significance, rendering it especially valuable for drug screening purposes [59, 292]. Therefore, multiple types of cells deposited during bioprinting are becoming a widely adopted approach. Most of co-culture models were bioprinted via EBB or DBB, enabling the study of interactions between diverse cell types [18, 101, 293]. The bioprinting technology has the capability to create models with cell composition, complex 3D micro-structures, biomaterial properties and vascularization that closely resemble physiological conditions [294, 295]. By integrating microfluidic devices, functional maturation and maintenance can be better supported, particularly for large-scale vascularized tissue constructs [296, 297]. With the progress made in these areas, the ultimate objective of creating a human-on-a-chip can be accomplished, leading to a comprehensively unified system capable of investigating the interconnected effects of various miniature organs. This integration, achieved through microfluidics networks, has the potential to revolutionize future drug testing methodologies [298, 299].

Post-fabrication processing is also important in biofabrication. Bioreactors will help the maturation of fabricated tissue, and technological advancements in imaging systems, biosensors, and analytical tools will create new possibilities for utilizing bioengineered 3D models in drug discovery endeavors [4, 144, 284]. Despite respectable advancements achieved from one or a few features, difficulties persist in effectively integrating all physiologically-significant elements to replicate a specific microenvironment. In the future, progress in biofabrication technologies, advancements in materials, and improved cell sourcing will enable the creation of a microenvironment with accurate and essential physical, chemical, and biological properties. Finally, in order to fully unlock the potential of 3D disease models for precise medicine, there is a need for continuous progress in technology and research across the realms of medicine, biology, and engineering. These advancements will address the existing challenges and pave the way for transformative outcomes.

5. Intellectual Property, Industry, and Regulatory Landscape

Although 3D models have shown the capacity to serve as a feasible alternative to animal models in the realm of drug discovery and advancement, it can be imagined that researchers may exhibit a certain level of hesitancy towards substituting animal models in the drug development process, which demands the cooperative endeavors of researcher, industry, regulatory departments. Among biofabrication technologies, bioprinting is attractive and promising due to its potential for automation and standardized processing. One recent report analyzed 176 bioprinting-related patents from 2006 to 2019 [300], a significant emphasis was placed on 3D tissue models within these patents. Among the various technologies, the largest group consisted of 50 patents related to 3D models, accounting for 28.4% of the total. Within this group, the most common applications were artificial tissues/organs (41%; 21 patents) and bioprinting of cells, tissues, and organs (31%; 16 patents) bioprinting. The artificial tissues/organs category mainly found applications in industrial and research settings, with drug testing, research involving disease models and toxicity assessment being the most cited applications. However, the absence of verified scientific validation and the challenges posed by regulations were identified as important obstacles in the application of bioprinted human tissues for advancing drug discovery and development. The scientific validation necessary for regulatory approval of new technologies can be achieved through testing with bioprinted models [301]. If those models are reliable predictors of human toxicity, their acknowledgement by the scientific community, which should naturally lead to subsequent regulatory adaptations [302]. Overall, researchers should actively seek and gather ample data from various experiments to enable the utilization of bioprinted models with increased confidence. As a result, over time, these models are likely to gain acceptance from regulatory departments because of the increasing amount of supporting evidence. However, during the initial phases of their implementation in laboratories, they can only serve as complementary to animal testing. In addition to further industrialization processes, sample consistency should be particularly important for standard drug screening protocols and reliable results [301]. 3D bioprinted models should be highly consistent in structure, function, and maturation level. Overall, advancements in 3D bioprinting and medical research are required to overcome present obstacles. Therefore, getting ready for what lies ahead is essential for both researchers engaged in advancing knowledge, technology, industries, and governments supporting and financing those activities.

6. Conclusions

The progress and utilization of 3D tissue models in preclinical applications is still in its nascent phases of advancement and the increasing adoption of 3D models presents the opportunity to efficiently assess the appropriateness of drug candidates for subsequent preclinical trials, providing quicker outcomes and cost-efficiency. The continuous advancement of innovative technologies strives to create 3D models that accurately emulate the in-vivo environment on a large scale. This ensures their suitability for downstream analysis techniques crucial for drug screening purposes. From simple spheroids to organoids, 3D models move forward to the combination of microfluidic chips and bioprinting heterocellular constructs in a dynamic environment within tuned biomaterials to accompany real-time analysis. Personalized therapeutics has been established and widely recognized as one of the most accepted prospective applications. Considering the complex structure of 3D models and their capacity to replicate the in-vivo surroundings, it becomes essential to standardize experimental protocols for drug screening and meet quantification criteria. Sufficient data from a large number of studies should obtain scientific validation. Consequently, regulatory agencies and pharmaceutical companies are likely to naturally embrace 3D models as an essential tool to strengthen and streamline the drug discovery process.