Engineered Microsystems for Spheroid and Organoid Studies

Abstract

Three-dimensional (3D) in vitro model systems such as spheroids and organoids provide an opportunity to extend our physiological understanding using recapitulated tissues that mimic physiological characteristics of in vivo microenvironments. Unlike two-dimensional (2D) systems, 3D in vitro systems can bridge the gap between inadequate 2D cultures and the in vivo environments, providing novel insights on complex physiological mechanisms at various scales of organization, ranging from the cellular, tissue-, to organ-levels. To satisfy the ever-increasing need for highly complex and sophisticated systems, many 3D in vitro models with advanced microengineering techniques have been developed to answer diverse physiological questions. This review summarizes recent advances in engineered microsystems for the development of 3D in vitro model systems. We highlight and discuss in detail the relationship between the underlying physics behind the microengineering techniques, and their ability to recapitulate distinct 3D cellular structures and functions of diverse types of tissues and organs. We also introduce a number of 3D in vitro models and their engineering principles. Finally, we summarize current limitations, and provide perspectives for future directions in guiding the development of 3D in vitro model systems using microengineering techniques.

1. Introduction

Research with well-defined yet more physiological three-dimensional (3D) in vitro models help gain a better understanding of complex in vivo biological systems.^[1-3] 3D in vitro models have emerged as a promising model system for mimicking of in vivo microenvironments, where complex cell-cell and cell-extracellular matrix (ECM) interactions occur.^[4,5] Conventional adherent two-dimensional (2D) monolayer cell cultures provide a simple, robust, and low-cost way to investigate biological phenomena with high reproducibility. However, 2D culture systems have many inherent limitations, including the restricted interactions between the cellular and extracellular environments, lack of tissue architectures, non-physiological structural and morphological characteristics, and an inability to represent complex cellular heterogeneity. As a consequence, many important biological processes are not adequately modeled in the 2D cultures. Therefore, many 3D culture systems have been proposed to alleviate such problems.

3D in vitro models can be classified into three major structure types based on their complexity: spheroids, multicellular spheroids, and organoids (Figure 1). These terminologies are often ill defined, and even used fairly inconsistently amongst researchers, and therefore need to be clearly established in the community. In general, the three structure types are defined based on the origin (i.e., immortalized cell lines, primary cells, or stem cells), number of cell types, model environment, complexity of the structure, and organotypic phenotypes/characteristics.^[6-8] More specifically, spheroids in this article refer to aggregated 3D structures created in a scaffold-free environment made from a broad range of cell types. Multicellular spheroids refer to the spheroids generated from multiple cell types. Spheroids may or may not exhibit polarity, and therefore can be limited in mimicking the composition and functionality of tissues/organs. Organoids, rather more controversially defined, refer to 3D structures generated from pluripotent stem cells and/or organ progenitor cells. Organoids are typically cultured in a scaffold such as basement membrane extract (BME) or ECM to provide a closer-to-ideal growth environment. A key difference that distinguishes organoids from spheroids is that organoids often better represent the cellular heterogeneity and physiological functionality of organs. In short, these 3D structures can exhibit the organ-specific structures and functions of diverse cell types, which is not achievable in 2D cultures where there exists only limited cell-cell and cell-matrix interactions.^[9] Therefore, the development of platforms for generating 3D structures has become highly important for achieving in vitro physiological microenvironments to recapitulate the structures and functions of tissues and organs.

Figure 1.

Schematic representation of 3D cellular structures: spheroid, multicellular spheroid, and organoid. Spheroids are compact aggregates of cells, whereas organoids show realistic microanatomy and typically have a hollow lumen. These 3D cellular structures are important basis for in vitro models of development, disease, drug design. The organization of the 3D cellular structures is composed of three different compartments: a necrotic core in the center, a quiescent zone of non-proliferating cells, and an outermost layer with proliferating cells. Different substance-dependent concentration gradients exist in these layers.

In vivo structures are generated via cell self-assembly processes that mimic morphogenesis, embryogenesis, and organogenesis.^[10] One example is the development of bone, in which mesenchymal progenitor cells undergo condensation resulting in tight cell–cell junctions by expressing adhesion molecules such as fibronectin, tenascin, syndecan, N-CAM, and N-cadherin.^[11-13] Although 3D in vitro models cannot fully reflect the mechanisms of such complex cellular interactions, they do mimic mesenchymal condensation to a certain extent.^[14] In addition, many studies have demonstrated the recapitulation of essential in vivo events including cell growth, differentiation, and morphogenesis by investigating the formation of 3D in vitro models in various physiological environments.^[15-18] In general, the dynamics of 3D structure formation follows three stages: 1) initial cell–cell contacts by cadherin–cadherin interaction and integrin–ECM protein binding, 2) delay period of cadherin accumulation for cellular reorganization, and 3) aggregate compaction by strong interactions of cadherins.^[19] However, this process may vary depending on different cell types, as the expressions of the proteins differ in cells. In addition, a simple model cannot solely describe the development of complex structures in vivo, as it merely describes an overall sequence of cellular events during in vitro cell aggregation.^[20] Therefore, more advanced and physiologically relevant in vitro models are needed to study the complex in vivo phenomena.

Various engineering methods have been developed to create 3D in vitro models, in which the cellular assembly mechanisms can be described by two principles: clustering-based self-assembly, and collision-based assembly.^[21] The clustering-based self-assembly occurs under static conditions, where cells sediment to the bottom of a liquid interface or low-attachment surface. Examples include hanging drop (HD),^[22-25] liquid overlay,^[26-29] microtiter plate,^[30-32] and thermal lifting.^[33-35] The HD method uses a small volume of cell suspension, and gravity-induced sedimentation that results in the spontaneous self-assembly of cells. The liquid overlay procedure promotes the self-assembly processes of the cells by preventing their attachment to the bottom of the plate. To achieve this, a cell suspension is loaded on a flat substrate made of low-adhesive materials such as agarose,^[36] polydimethylsiloxane (PDMS),^[37] poly(2-hydroxethyl methacrylate),^[29] or polyethylene glycol (PEG) hydrogels.^[38] Additionally, the materials can be engineered to form U-shaped concave structures using capillary force, which can enhance cell-to-cell interactions. In a slightly different approach, efficient cells aggregation has been achieved by applying external forces, such as electric fields,^[39] magnetism,^[40] and ultrasound waves.^[41] In contrast to clustering based self-assembly, collision-based assembly using fluid flow involves dynamic processes, wherein cells are made to contact each other by initiating collisions using, for example, centrifugation,^[15,42,43] spinner flasks,^[44-46] rotating wall vessel (RWV),^[47-49] or stirred tank reactors (STR).^[50-52] For instance, in so-called pellet cultures, 3D structure formation is achieved by centrifugation, which causes cell-to-cell collisions at the bottom of tubes. This method can rapidly form cell aggregates, and therefore offers a significant advantage in terms of the ease of operation. In addition, mixing-based techniques (i.e., spinner flask, RWV, and STR) promote cell-cell collisions with constant stirring, and are compatible with large-scale production processes. Despite the successful implementations of such 3D in vitro model systems, there are still many issues that need to be further addressed, such as a lack of complexity, need of greater physiological relevance, polydispersity, co-culture, and real-time monitoring.

Microengineering of 3D in vitro model systems are a new and innovative technique for providing more physiologically relevant and efficient recapitulation of complex in vivo microenvironments in a miniaturized system.^[53-55] In short, these platforms can not only provide more reliable and predictive preclinical results than the traditional 2D monolayer cultures, but also replace conventional methods of 3D cell culture, while also offering significantly more cost-effective solutions compared to animal models.^[56] Another important advantage is that these systems can facilitate personalized medicine applications that use various types of patient cells. Remarkable developments have been achieved over the past decade that opened up opportunities to study structural and functional characteristics of tissues and organs, and examples of the tissues/organs-of-interest range from the brain,^[57,58] liver,^[59,60] intestine,^[61-63] lung,^[64,65] stomach,^[66] kidney,^[67] heart,^[68-70] and retina.^[71]

Herein, we review advances in state-of-the-art microengineering-based 3D in vitro systems, with an aim to introduce engineers and biologists to relevant information and details. In particular, this review aims to (1) help biologists understand the operational mechanisms of current microengineering platforms, and (2) guide engineers in the innovation and improvements of these platforms. We categorize types of engineered microsystems, and summarize representative examples of their biological applications. In addition, we describe and explain the underlying engineering principles of microengineering techniques that play a critical role in creating various physiological conditions (Figure 2). Engineered microsystems utilize different types of forces including gravitational force, capillary force, centrifugal force, adhesive force, and hydrodynamic shear stress to provide innovative 3D culture platforms. These forces can affect organisms at the molecular, cellular, and complex structural levels. Examples include, but are not limited to, intrinsic adhesive forces in cell-cell and cell-matrix interactions, internal capillarity and shear force in vascular systems, and gravity. These forces influence both physical and biological phenomena, impacting development, homeostasis, and evolution of living organisms.^[72] In addition, reduced and increased gravitational forces induce physiological changes such as weight-bearing structures,^[73] musculoskeletal tissue fastening,^[74] enhancement of myogenic differentiation and myotube formation^[75]. Hydrodynamic shear stress is a frictional drag force imposed by blood flow that remains below ~70 dyn/cm² in normal vessels.^[76-78] This shear stress regulates the development of blood capillaries by triggering cadherin and integrin receptor-medicated cell-cell and cell-matrix interactions.^[79] The review concludes with discussions on desired improvements to address current limitations of microengineering-based 3D in vitro models, and perspectives on future applications such as disease modeling, tissue engineering, and clinical studies.

Figure 2.

Mechanical principles for development of in vivo and 3D in vitro model system. The five mechanical principles include gravitational force (F_g), capillary force (F_c), centrifugal force (F_w), adhesive force (F_ad), and hydrodynamic shear stress (τ_s), each of which can uniquely influence the formation of 3D cellular structures.

2. Engineered microsystems for spheroids and organoid culture

A wide variety of microscale techniques, hereafter also referred to as microengineering, have been used to perform 3D cellular engineering to mimic diverse physiological conditions. 3D in vitro model systems mimic specific microstructures and biochemical cues present in in vivo microenvironments.^[80,81] Engineered microsystems provide great versatility in generating in vitro models with homo- and heterogeneous cell types, controllable sizes, and configurable structural forms. Therefore, microengineering can provide well-defined functions and characteristics of organs or tissues.^[82-84] In particular, the choice of an appropriate platform for generating 3D cellular structures is mainly dependent on the cell types and their characteristic features.

This section describes the development of the different types of engineered microsystems to date, with specific emphasis on their respective fundamental physics, operating procedures, advantages and limitations for the formation of 3D cellular structures. These engineered microsystems can be classified into four categories as illustrated in Figure 3: microwell-, centrifugation-, liquid drop-, and microchannel-based cell culture system. These engineered microsystems have been used to culture 3D cellular structures of diverse origins, including pluripotent stem cells, embryonic or adult stem cells, primary cells, and cell lines.^[6,85,86]

Figure 3.

Schematic representation of engineered microsystems for 3D in vitro models. Diverse cell sources include, but are not limited to, primary cells, embryonic stem cells, adult stem cells, and induced pluripotent stem cells. Different engineered microsystems with distinct underlying mechanisms are used to generate 3D cellular structures with specific morphological and physiological characteristics.

2.1. Microwell cultures

A microwell system consists of an array of microscale wells or geometrically controlled microstructures patterned in a wide range of polymeric substrates such as polyester (PE),^[87] polyurethane (PU),^[88] polyethylene glycol (PEG),^[89-91] and many others.^[92-94] Microwell structures are designed to promote cell aggregation within a short period of time, resulting in the formation of 3D cellular structures with uniform sizes and shapes.^[95-97] The basic mechanism of this system involves loading of a cell suspension into the microwells, after which the cells are sedimented by gravity, thereby gradually initiating cell aggregation.

In general, fabrication of engineered microsystems involves lithographic processes. Some of the representative examples of such methods include photolithography,^[98] soft lithography,^[99] and nanoimprint lithography,^[100] each of which offers sets of advantages and disadvantages.^[101] First, photolithographic microfabrication uses photosensitive materials and optical radiations to transfer microscale patterns. Photolithographic methods allow precisely defined geometries on the micro-meter scale, providing excellent spatial control for 3D culture microenvironments. Second, soft lithography has achieved one of the most significant breakthroughs in microfabrication and is perhaps the most universally used method in academic research. Owing to the versatility of soft lithography, researchers have introduced many different techniques, and the list is ever-growing.^[102-105] Many engineered microsystems made with soft-lithography-based microcontact printing have demonstrated efficient recapitulation of in vivo microenvironments in different types of cellular systems.^[106-110] Lastly, it has been fairly recent for researchers to start utilizing nanoimprint lithography (NIL) for 3D culture systems, first introduced in 1996.^[100] The fundamental idea is to provide topographical nanostructured scaffolds that accurately mimic the native 3D microenvironments. NIL has been applied to a wide of range of cell types. Key examples of such applications include fibroblasts,^[111,112] osteoblasts,^[113] neurons,^[114-116] and MSCs.^[117] It is worthwhile mentioning that there are many important nonlithographic micropatterning methods such as microfluidic patterning, laminar flow patterning, stencil patterning, and emulsion freeze-drying.

Soft lithography starts with conventional photolithography where UV light exposure of photosensitive materials creates patterned photoresist structures. An innovation in soft lithography is to directly use these photoresist structures as the master mold to create replicas in silicone elastomers. Traditionally, this method has been most suitable for single-layered microstructures, but researchers have expanded it to create multi-layered structures through additional steps or repetition of specific steps.^[118] Microwells with flat-bottom surface have been efficiently fabricated using soft-lithography. Flat-bottom microwells have shown good performance in trapping cells and forming large quantities of uniform-sized spheroids, however, there are still some limitations. Because the bottom is flat, seeded cells may be distributed sporadically throughout the bottom surface, which may result in the formation of two or more spheroids in one microwell.^[119] This may lead to insufficient results in terms of uniformity of spheroids when compared to the spheroids produced in microwells with a hemispherical bottom.^[120-123]

To generate 3D cellular structures, a cell suspension with essential nutrients is loaded into biocompatible substrates, such as hydrogels, plastics, or PDMS. For example, Casey et al. have demonstrated the use of hydrogel-based microwells for efficient generation of 3D cellular structures using HCC1806 human breast cancer cell line (Figure 4a).^[124] The microwells consisted of PEG-acrylate, a non-biofouling polymer that shows favorable hydration property thereby restricting cell adhesion. The HCC1806 cells were cultured for up to seven days in the microwells, from which the spheroids were collected. However, the hydrogel-based microwell systems have a fundamental problem in maintaining their structural integrity, owing to swelling properties that inherently limit long-term 3D cell cultures. To address this drawback, Chao et al. developed simple-to-use microwells that are made by assembling a bottomless PDMS plate and an agarose hydrogel sheet (Figure 4b).^[125] This device exhibits a nontoxic and biocompatible feature with reduced deformability, and therefore provides long-term culture capability. In addition, the agarose hydrogels not only provide non-adherent environments for enhanced cell aggregation, but the material properties can also be easily engineered to control the 3D microwell environments to manipulate the multicellular spheroid formation.^[126,127] Using the PDMS/agarose-based microwells, they demonstrated the successful formation of hepatocyte spheroids (HepG2), with which they studied the relationship between the spheroid formation and culture conditions such as seeding density and microwell diameters. Bao et al. devised a method of fabricating methacrylated hyaluronic acid (MeHA) hydrogel microwells with various shapes^[128] and controllable stiffness.^[129] However, the system was not used specifically for 3D culture applications. Nevertheless, the advanced microwells were utilized to culture hMSCs, which showed that mechanical properties of cells were affected such as actin distribution, focal adhesions, nuclear shape, yes-associated protein/transcriptional co-activator with PDZ-binding motif (YAP/TAZ) localization, cell contractility, nuclear accumulation, differentiation, and stress fiber organization. Similarly, it was shown that microwell shape caused changes in wound healing processes in NIH T3T fibroblasts.^[130] Despite the successful implementation of hydrogel-assisted microwells for the generation of 3D cellular structures, this method still faces many challenges, and therefore needs further improvements. In particular, most, if not all, hydrogel-based microwells have a flat bottom surface, onto which cells randomly sediment, causing the formation of multiple cell structures per well. In addition, there are only a limited number of hydrogels with ultra-low attachment (ULA) properties, and the development of non-adherent materials with specific microstructures is required.

Figure 4.

Microwell platforms with various microstructures: a) Schematic illustration for the fabrication of the microwell using PEG hydrogels (20 kDa). Reproduced with permission.^[124] Copyright 2017, IOP Publishing. b) Agarose microwell array with improved throughput of hepatocyte spheroid formation. Reproduced with permission.^[125] Copyright 2020, ACS Publications. c) U-shaped polyHEMA microwell for enhanced T47D cell aggregation and sphere formation. The scale bar is 100 μm. Reproduced with permission.^[131] Copyright 2015, Springer Nature Publishing AG. d) Aggregation of bone marrow-derived mesenchymal stem cells in the microwell-mesh and the growth of the cartilage microtissues at day 14 post seeding. The microtissues were too large to pass back through the nylon mesh and were retained in discreet microwells. Reproduced with permission.^[134] Copyright 2015, Elsevier.

Microwells with different geometries have been developed to promote enhanced cell aggregation and generation of 3D cellular structures.^[93] One promising strategy is to modify the microwell designs to have U-shaped microstructures for improved cell–cell interaction as shown in Figure 4c. For example, Chen et al. proposed U-shaped PDMS microstructures as a high-throughput system for cancer spheroid formation.^[131] The PDMS materials exhibit eight key properties that have made PDMS the material of choice for rapid prototyping of microwell systems: 1) ease of molding, 2) elastomericity, 3) biocompatibility, 4) optical transparency, 5) gas-permeability, 6) non-toxic, and 7) freedom from intellectual property restrictions.^[132] To make microwells with a U-shaped bottom, a mixture of poly(2-hydroxyethyl methacrylate) (polyHEMA) and ethanol was applied on a plasma-treated PDMS substrate to increase the hydrophilicity of PDMS, resulting in the adhesion of the polyHEMA to the patterned PDMS substrate during a stamping process. As the ethanol evaporates from the mixture, the polyHEMA solidifies, forming a concave meniscus that gets coated onto the PDMS microwells. Using the modified U-shaped microwells, Chen et al. demonstrated highly reproducible and long-term (> 2 weeks) spheroid cultures of diverse breast cancer cell lines including T-47D, MCF-7, and SUM-159 cells with low spheroid size variance (< 10%). Moreover, Mori et al. demonstrated the culture of liver cell organoids using microwells.^[133] The culture chip comprises cell adhesive and non-adhesive areas created by controlling the ratio of collagen and PEG, respectively. This chip can generate uniform HepG2 cell organoids with liver functions such as protein secretion and ammonia removal. A limitation of the microwell system is the possible loss of biological replicates during required medium exchange procedures. Therefore, Futrega et al. developed a microwell-mesh system where a nylon mesh covers over the microwell openings, providing discrete compartmentalization for the cells (Figure 4d).^[134] The microwell was fabricated by using PDMS soft-lithographic replication from a polystyrene negative master mold. With the microwell–mesh platform, they showed that 3D microstructures of bone marrow-derived mesenchymal stem cells (MSCs) were larger than the mesh pore size, and therefore were efficiently maintained within the discrete microwells. In addition, this strategy enabled a longer and more stable culture for over 21 days, even with several medium exchange iterations. Brandenberg et al. recently produced PDMS-imprinted hydrogel-based U-shaped microwells with a convex micropillar structure to culture mouse and human gastrointestinal organoids.^[135] Their advanced system was capable of performing automated in situ hydrogel equilibration, cell seeding, medium changes, drug exposure, and statistical data analysis. In summary, recent advanced microwell techniques have shown the ability to generate multicellular structures with uniform and compact 3D structure in a high throughput manner. Nevertheless, there is still a room for improvements such as regulating cell-to-cell and cell-to-matrix interactions in the microwells beyond what is possible with passive processes. Some recent advances in the development of functionalized microwell platforms have demonstrated the capability of achieving complex 3D cell culture systems.^[136-138]

2.2. Centrifugation-based culture

Centrifugal force is an inertial force on objects rotating at a constant angular velocity that is directed radially outward from the center of rotation. Researchers have utilized this principle in microfluidics to implement efficient mixing, valving, and separation for diverse biological applications.^[139-141] Centrifugal force has also been used in cell biology for the formation of 3D cellular structures. To generate 3D cellular spheroids, cells must be aggregated with high cell densities; and centrifugal force can play a significant role as it can generate forces over hundreds of times that of gravity. Therefore, centrifuge-based 3D cell culture systems have received a great amount of interest owing to advantage of being able to rapidly promote high-density cell aggregation. Research on using centrifugal force in the formation of 3D cellular structures has evolved into the forms of pallet culture, rotary culture, centrifugation-assisted microwell culture, and lab-on-a-compact-disc (CD) culture systems.

Pellet culture refers to the aggregation of cells as “pellets” by centrifugation, hence inducing the formation of 3D cellular structure.^[142,143] Pellet culture has been widely used in the study of single spheroid generation, owing to the simple principle of operation.^[144,145] However, this method has an inherent disadvantage of low throughput, as only one spheroid can be generated per tube.^[143] Therefore, although pellet culture systems are still frequently used today, they are not preferred for high-throughput studies.

The rotary culture is a method whereby cell aggregation is promoted in a media suspension under the influence of constant rotation of a rotary shaker, leading to generation of a large number of 3D cellular structures.^[146,147] An early study showed efficient rat hepatocyte spheroid formation by rotating a culture dish. Using a similar approach, Lei et al. demonstrated the formation of embryoid body (EB) from mouse embryonic stem cells (ESCs) (Figure 5a).^[148] They compared the degree of differentiation of ESCs into EB in a static culture environment, and with a rotary culture environment at 20 RPM. As a result, it was confirmed that a significantly larger percentage of cystic EB was generated in rotary culture (55.2%) than in static culture (25%). Unlike pallet culture, rotary culture can generate more than 10,000 spheroids at once. However, the spheroids obtained through rotary culture are often irregular in shape, and have a broad size distribution.^[149] To compensate for this limitation, microwell centrifugation has been introduced.^[150] As discussed earlier, microwells alone can generate 3D cellular structures; however, the aggregation time, efficiency, and size uniformity can further be improved by using centrifugation.^[131] One study successfully generated human glioblastoma (U87MG) spheroids in a round-bottom 384-well plate to make a tumor cell drug screening model.^[151] The spheroid sizes (100–600 μm) were precisely controlled through the number of cells seeded. The authors tested and compared the efficacy of an anticancer drug (7-N-Allylamino-17-demethoxygeldanamycin, or 17-AAG) in 2D and 3D models. Another study demonstrated multicellular organoids of islet cells (ICs) and human amniotic epithelial cells (hAECs) to be capable of insulin synthesis and secretion (Figure 5b).^[152] ICs and hAECs at a 1:1 ratio (500 cells/organoid) were injected into 3D agarose-patterned microwells of diameter 400 μm. The authors demonstrated that the generated IC-hAEC organoids showed higher insulin secretion in both normoxic and hypoxic conditions compared to the IC-alone spheroids. Another recent paper reported multicellular testicular organoids from porcine, murine, human, and primate testicular cells cultured in centrifugation-assisted microwells.^[153] Different types of cells were self-organized in distinct compartments; germ cells and Sertoli cells were located outside the collagen IV basement membrane, peritubular myoid cells were located inside the membrane, and Leydig and endothelial cells were located at the core of the interior compartment. The multicellular testicular organoids, which had the cellular compartments similar to a natural testicle tissue, showed reduced retinoic acid response and lower level of autophagy compared to the 2D model. To summarize, the microwell centrifugation systems offer a very efficient method of generating 3D cellular structures their fast aggregation speed, high-throughput, and stability. Recently, it has been shown that lab-on-a-CD systems can efficiently generate 3D cellular structures.^[154,155] The principle of lab-on-a-CD systems involves fluid flow through rotation inside a compact disc-type chip. The Lab-on-a-CD systems offer a significant advantage as they offer the capability of 3D structure formation and drug screening simultaneously in a fluid-flowing chip. One study has demonstrated the formation of multicellular spheroids using human adipose-derived stem cells (hASC) and human lung fibroblasts (MRC-5) on a lab-on-a-CD chip (Figure 5c).^[154] In this study, multicellular spheroids of three different types, sandwich, concentric, and Janus, were generated by manipulating the number of cells and the injection interval. The spheroids generated in this system showed larger diameter with better sphericity than spheroids generated in conventional microwell systems. Another study demonstrated lab-on-a-CD-based lung cancer organoid (LCO) culture for drug sensitivity testing.^[155] The organoids were cultured from cells extracted from small-cell lung cancer patients. The 3D culture process involved embedding of the extracted cells, and injection of the resulting suspension to the chip for drug testing. Centrifugation of the device caused formation of LCOs with diameters ranging from 120 to 180 μm, and the generated LCOs were tested with common-used anti-SCLC drugs such as cisplatin and etoposide in flowing conditions. However, the research on using lab-on-a-CD systems for spheroid/organoid applications is still in its early phase, and therefore needs further studies. One of the very recent papers has demonstrated a centrifugation-assisted droplet-based 3D culture system that is capable of generating a large amount of encapsulated cell spheroids within a customized centrifuge device (Figure 5d).^[156] The nozzle was made by inserting a glass capillary device into a blunt-end needle, which was then assembled with a conical tube for the custom device. The centrifuge device containing a suspension of alginate hydrogels and human malignant melanoma cell line cells (MEL28) were centrifuged at 332 g (1250 RPM). Upon centrifugation, the alginate hydrogels extruding from the nozzle formed uniform droplets. The generated droplets were then crosslinked as soon as they made contact with aqueous calcium solution at the bottom of the tube, producing alginate microspheres encapsulating the cells. They demonstrated that the encapsulated spheroid cell densities can be controlled by seeding cell counts, and obtained an optimal condition of 400 cells/droplet. In short, this new method utilized centrifugal force to conduct scaffold-supported 3D cell cultures, significantly improving the production throughput and uniformity of templated cell culture populations for suspension culture or large batch bioreactor studies.

Figure 5.

Centrifuge-based 3D cellular structures culture systems: a) rotary culture, where cells are constantly rotated to prevent sediment, and a large number of spheroids are generated in the media. Mouse embryonic stem cells (mESCs) were formed into embryoid bodies (EBs) using rotary culture. Reproduced with permission.^[148] Copyright 2014, Springer US. b) microwell culture of islet cells (ICs, green) and human amniotic epithelial cells (hAECs, red) assisted by centrifugal force. It is essentially the same principle as the microwell platform, but by applying centrifugal force, the denser spheroid is generated faster. Scale bar is 50 μm. Reproduced with permission.^[152] Copyright 2019 Nature Publishing AG. c) Lab-on-a-CD culture, where a disc-shaped chip is rotated to generate various types of multicellular spheroids. Multicellular spheroids of human adipose-derived stem cells (hASC, green) and human lung fibroblasts (MRC-5, red) were generated. Scale bar is 500 μm. Reproduced with permission.^[154] Copyright 2017, IOP Publishing. d) Schematic of centrifugal droplet-generating device and photomicrograph of human malignant melanoma cell line (MEL28) encapsulating microspheres cultured for a time course of 7 days. Scale bar is 200 μm. Reproduced with permission.^[156] Copyright 2020, SAGE Publications.

In summary, centrifuge-based systems have been used for a long time owing to their ease of operation, and the advantage of rapid formation of 3D cellular structures. Diverse engineering methods of centrifugation have been introduced for efficient generation of multicellular spheroids and organoids. However, it still remains unclear as to how centrifugation, which does not mimic physiological conditions, can sometimes lead to efficient formations of spheroids and organoids.^[157] It is obvious that an intense centrifugal force applied for a prolonged period of time to cells is not desirable. Therefore, in order for the centrifuge-based systems to be more actively utilized as efficient cell culture systems for 3D in vitro models, more in-depth studies on the underlying biological basis need to be conducted.

2.3. Liquid drop-assisted culture

With the high controllability of microfluidic techniques, there have been significant improvements in the efficiency and uniformity of 3D cellular structures, providing better high-throughput in vitro models with minimal human intervention.^[25,158,159] Among many microfluidic techniques, encapsulation of cells in a liquid template, also called an “organoids-in-a-drop” system, has gained popularity owing to its advantages in terms of high cell viability inside innate liquid phase, and ease of manipulations.^[160-162] Typically, there are two types of liquid drop-assisted culture systems: static HD platforms^[163-165] and dynamic emulsion-generating microfluidic devices.^[166-168]

A number of techniques have been developed that can enable automated, high-throughput, and scalable processing of parallelized 3D cell culture. In an early work by Tung et al., a novel technique was proposed in which the use of automated liquid handling robots significantly reduced labor costs compared with manual pipetting procedures.^[25] Moreover, Wu et al. developed so-called a micro-hanging drop (μHD) platform, which can address the drawbacks of the conventional HD method, such as labor-intensive procedures, limited droplet volume, short-lived culture, and difficulty of media change (Figure 6a).^[169] This platform had automation capabilities, providing facile ESCs loading, efficient medium exchange, computerized monitoring, and intact recovery of 3D spheroids without pipette-based manual procedures. The cells, docked in each opening hole by hydrostatic pressure-driven flow, were then allowed to settle to the droplet bottom, leading to the formation of hanging drops; in each, the cells aggregated to develop 3D structures. In addition, the computerized medium exchange can be performed with the pumps to continuously maintain long-term 3D culture of the ESCs (>10 days). In a different approach, researchers used HD array systems to generate fluidic interconnections to enable cellular communication among different cell types, thereby providing potential in vitro models of multiorgan systems. For instance, Frey et al. proposed a fluidically-interconnected HD device composed of rim structures by connecting open conduits (Figure 6b).^[170] The complete open structure allows efficient gas exchange, and prevents bubble formation during incubation. The cell suspension was first loaded by manual pipetting, after which the solution moved along the fluidic channels via capillary force, eventually leading to the formation of equally-sized hanging drops. High-throughput formations of human colorectal carcinoma (HCT-116) spheroids have been achieved in the HD microfluidic networks under a continuous flow of fresh medium. They investigated complex multi-tissue interactions by using capillary valving that can modulate cellular crosstalk between neighboring columns. Two different types of spheroids derived from rat liver (rLiMT) and cancer (HCT-116) cell lines were cultured in the same device to study anticancer effects by hepatic activation. Perfusion of cyclophosphamide (CP), which is a prodrug that gains an anticancer effect upon hepatic activation,^[171] was shown to significantly reduce the growth of the tumor spheroids in the rLiMT-HCT116 co-culture system, compared to the HCT116 spheroid monoculture. Although the HD platform with automation and miniaturization is a promising approach for engineering of self-assembling 3D cellular structures, the lack of biologic scaffold in the fluid environments is an important issue to be addressed. Therefore, incorporating 3D cellular scaffold materials, such as hydrogels, ECM, and other types of engineered matrices, could be one way to develop more complex physiological environments, hence providing a means of identifying essential biophysical and biochemical signals of the cell-substrate interactions.^[172]

Figure 6.

Liquid drop-assisted devices with different manipulations of cell suspension: a) Microfluidic hanging drop (μHD) chip operation and docked embryonic stem cells aggregated at bottom center of hanging droplets. The scale bar is 200 μm. Reproduced with permission.^[169] Copyright 2016, MDPI. b) Cross-section of channel structures defined by rims creating uniform-sized interconnected hanging drops (~14 μl). Reproduced with permission.^[170] Copyright 2014 Springer Nature Publishing AG. c) Emulsion-generating microfluidic device for fabrication of mesenchymal stem cells (BMSCs)-laden GelMA microspheres. Scale bars are 100 μm. Reproduced with permission.^[167] Copyright 2016, John Wiley & Sons. d) Electrospray-based device for 3D cell aggregation between MDA-MB-231 (with red fluorescence) and MCF10A cells. Reproduced with permission.^[180] Copyright 2015, The Royal Society.

Microfluidics-assisted dynamic droplet generation can be an alternative means for the 3D in vitro model development. These platforms achieve high-throughput with well-defined in-vivo-like microenvironments that not only promote enhanced cells interactions, but also provide physiologically relevant mechanical and biochemical cues for the evolution of 3D structures with specific functions. Moreover, another significant advantage of such platforms is the scalable integration by serial connections of microfluidic channels for more efficient manipulation of the 3D culture systems.^[173-175] Hydrogels have been extensively applied in conjunction with microfluidic technology for the microencapsulation of living cells, owing to their low toxicity, biocompatibility, and well-documented crosslinking chemistries.^[176] In addition, crosslinked hydrogels generate the porous networks, facilitating efficient exchange of nutrients, gases, and metabolic wastes, and control the mechanical properties such as stiffness, roughness, and biodegradability.^[177] For example, gelatin methacrylate (GelMA) microspheres encapsulating bone marrow-derived mesenchymal stem cells (BMSCs) are generated by a glass capillary microfluidic device (Figure 6c).^[167] Gelatin is a biopolymers obtained from partial hydrolysis of collagen that contains bioactive and cell adhesive sequences, leading to cell attachment, proliferation, and differentiation.^[178,179] The encapsulated cells show growth and proliferation within the GelMA microspheres with high cell viability of > 60% at seven days after encapsulation. They also confirmed that the temporal hallmarks for osteogenic development, such as alkaline phosphatase (ALP) expression and calcium deposition, to verify their in vitro system. 3D spheroids may exist not only in homogeneous organizations of one cell type, but also as heterogeneous assemblies consisting of various cell populations. Lu et al. investigated cell–cell interaction by encapsulating human mammary epithelial cells and human triple negative breast cancer cells (MCF10A and MDA-MB-231) within ECM-engineered alginate hydrogel microparticles (Figure 6d).^[180] Using a multifluidic electrostatic spraying device, they produced core-shell hydrogel microparticles composed of a Matrigel core and an alginate shell. A random mixture of the two cells assembled and organized themselves over time into a well-defined core-shell segregation, where the MCF10A cells enclosed the MDA-MB-231 owing to the difference in E-cadherin expression levels. Similarly, the authors also performed co-culture of primary rat hepatocytes and mouse 3T3-J2 fibroblast stromal cells in the microparticles. A significant improvement of hepatocyte viability was observed in the co-culture, which was evident from the upregulation of liver-specific genes such as cytochrome-P450 (CYP450) and phase II genes.^[181] Therefore, advances in μHD systems and microfluidics-assisted dynamic droplet platforms provide advantages in terms of degrees of freedom in precise control of the in vitro microenvironments for efficient optimization of the formation of 3D structures.

2.4. Microchannel-based culture system

3D cell culture can be performed in microstructures, which bridges the gap between 2D cell culture and organ-on-a-chip systems. This strategy provides a way to investigate the effects of mechanical stimuli and biochemical treatments on 3D in vitro cellular models in confined microstructures. For example, a hydraulic pressure-driven passive pumping method offers a simple and effective cell loading capability to perform uniform cell trapping and removal of spheroids.^[182] The system consists of horseshoe-shaped cell trapping barriers whose membrane pressure can be controlled to enable efficient cell trapping, growth, and extraction (Figure 7a). Successful 3D microstructure formation was demonstrated using human bronchoalveolar carcinoma cells (H1650), which were trapped and aggregated in the trapping barriers. In addition, the extraction of the generated 3D structures was demonstrated by gradually increasing the inlet pressure. While this system is capable of forming and extracting uniform 3D spheroids, it has a limitation of low throughput owing to the limited number of microstructures per devices. Therefore, researchers proposed ingenious designs for microfluidic devices with efficient trapping microstructures that offer high-throughput production. In a new study, Liu et al. demonstrated a pneumatic microfluidic device with highly-parallelized spatial cell trapping for 3D cell culture (Figure 7b).^[183] A pneumatic microfluidic system was developed to facilitate the 3D cell culture with simple cell loading, direct entrapment, and precise flow manipulation.^[184,185] Human glioma (U251) cells were mechanically trapped by activating pneumatic microstructures that were controlled by an external solenoid valve. It was reported that cell proliferation, viability, and behavior were dependent on various experimental factors such as flow rate, exogenous growth factors, and cell density. In addition, the pneumatic microfluidic device was able to maintain long-term (> 30 days) 3D tumor spheroids, while supporting operations in a dynamic, reproducible, and high-throughput manner.

Figure 7.

Microchannel-based cell culture system: a) Horseshoe-shaped barrier for cell trapping, clustering, spheroid formation and incubation of non-small lung cancer cells (H1650). Reproduced with permission.^[182] Copyright 2011, The Royal Society. b) Pneumatic microstructures (PμSs) for spatial cell trapping fluid flow and formation of 3D tumors (U251 cell). Reproduced with permission.^[183] Copyright 2015, ACS Publications. c) Enhanced vascularization of kidney organoids within a perfusable milli-fluidic system under varying fluidic shear stress. Reproduced with permission.^[186] Copyright 2019, Springer Nature Publishing AG. d) Electrodynamic digital microfluidic (DMF) fluid manipulation for HepG2 liver organoid culture. Reproduced with permission.^[188] Copyright 2014, The Royal Society.

Microfluidic platforms can also be engineered to mimic physiological environments, providing relevant biomechanical cues for developing 3D in vitro models. For example, a 3D-printed flow-assisted culture platform demonstrated structural and functional maturation of human induced pluripotent stem cells (hiPSCs) into kidney organoids (Figure 7c).^[186] The platform was subjected to configurable fluidic shear stress to investigate its effect on organoid formation. In particular, it was shown that shear stress ranging 0.008-0.035 dyn/cm² induced vascularization with enhanced expression of vascular markers such as MCAM and PECAM1 after 10 day of perfusion, while applying lower shear stress conditions (~0.0001 dyn/cm²), suggesting that shear stress is an important cue.^[187] Recent advances in digital microfluidics (DMF) have enabled potent integration into 3D cell culture, which offers diverse opportunities including, but not limited to, programmable microarray generation and independent manipulation by electrowetting-based fluid control. As shown in Figure 7d, an electrohydrodynamics-based DMF platform can conduct all-in-one on-chip 3D cell culture with no external equipment and valveless fluid control.^[188] The authors focused on the generation of a hepatic tissue model, and investigated the cellular characteristics such as contractility, viability, and albumin production by controlling individual experimental factors in the digital microfluidic platform. Human liver cancer cells (HepG2) were co-cultured with NIH/3T3 fibroblasts in different concentrations of collagen I, leading to the formation of 3D liver structures. In addition to the programmable controllability, the system also includes an on-chip monitoring tool that offers an efficient evaluation capability.

As described in the aforementioned advances, many engineered microchannel platforms have been developed to offer promising improvements such as long-term culture, precise control over many factors, dynamic handling of microenvironments, high-throughput capability, and automation of time-consuming procedures. These platforms have successfully generated well-defined 3D cellular models, which have been used in studying basic cell biology to disease pathophysiology.

3. Mechanical principles of engineered microsystems

As seen by the recent advances in the different types of microengineering techniques presented above, in vitro 3D models of diverse cell/tissue/organ types show significant potentials for pathophysiological and oncological studies, drug discovery, tissue implantation, and regenerative medicine development. In all microengineering systems, different mechanical principles provide the fundamental basis for the formation of 3D cellular structures (Table 1). More explicitly, these principles include gravitational force (F_g), capillary force (F_c), centrifugal force (F_w), adhesive force (F_ad), and hydrodynamic shear stress (τ_s). The mechanical principles influence the modulations of single or multiple cells and determine their cell fates, thereby providing desirable manipulations for the specific cell/tissue/organ-of-interest. The engineering of mechanical phenomena at the cellular level includes cell-cell and cell-ECM adhesion, rheology, cell mechanics, and mechanotransduction, which are all important to effectively mimic in 3D in vitro systems. At the macroscopic system level, appropriate microenvironments for the formation of 3D cellular structures can be created by sedimentation, aggregation, and entrapment using external forces. Therefore, we introduce the underlying physics and roles of the representative mechanical principles for 3D cell cultures in engineered microsystems.

Table 1.

Representative examples of microengineered platforms with mechanical principles to generate organoids and spheroids.

Target tissue	Platforms	Principle	Cell type	in vitro model	Ref.
Brain	Microwell	Fg, Fc, Fad	Prenatal rat cortical neurons	Spheroid	[254,255]
			ReNcell VM human neural stem cells	Spheroid	[256]
			Human embryonic stem cells	Organoid	[257,258]
			Human induced pluripotent stem cells	Organoid	[259]
Lung	Microwell	Fg, Fc, Fad	Primary normal human lung fibroblasts	Spheroid	[260]
Liver	Microwell	Fg, Fc, Fad	Rat hepatocytes	Organoid	[261]
			Human fetal liver cells	Organoid	[262]
	Microchannel	Fad, τ	Human hepatoma cell line and human umbilical vein endothelial cells	Organoid	[263]
			Human pluripotent stem cells	Organoid	[264]
			Primary human hepatocytes	Organoid	[186]
			Primary mouse embryonic fibroblast cells	Organoid	[265]
Heart	Microwell	Fg, Fc, Fad	Human and murine pluripotent stem cells	Organoid	[266]
			Human induced pluripotent stem cells	Organoid	[267]
Intestine	Microwell	Fg, Fc, Fad	Mouse crypts	Spheroid	[268]
				Organoid	[269]
	Centrifuge	Fw, Fad	Single intestinal stem cells	Organoid	[270]
	Liquid drop (μ-fluidics)	Fad	Human gut cell line (Caco-2)	Organoid	[271]
	Microchannel	Fad, τ	Intestinal epithelial cells	Spheroid	[272]
Kidney	Microwell	Fg, Fc, Fad	Human pluripotent stem cells	Organoid	[273]
Bone	Liquid drop (μ-fluidics)	Fad	Human bone marrow-derived stem cell	Spheroid	[168]
Pancreas	Liquid drop (μ-fluidics)	Fad	Human pancreatic endocrine cells	Organoid	[274]
Testicles	Centrifuge	Fw, Fad	Human testicular cells	Organoid	[153]
Ovary	Microwell	Fg, Fc, Fad	Human placenta-derived mesenchymal stem cell	Spheroid	[275]
Tumor	Microwell	Fg, Fc, Fad	Human breast adenocarcinoma	Spheroid	[276]
			Human lung adenocarcinoma	Spheroid	[277]
			Human ovarian cancer cells (SK-OV-3)	Spheroid	[278]
	Liquid drop (μ-HD)	Fg, Fad	Human colorectal carcinoma cells	Spheroid	[170]
	Liquid drop (μ-fluidics)	Fad	Breast tumor cells (MDA-MB-231, MCF-7, and SK-BR-3)	Spheroid	[279]
			Small-cell lung cancer	Organoid	[155]
	Microchannel	Fad, τ	Human oral squamous cancer	Spheroid	[280]

3.1. Gravitational force

Gravitational force (F_g) is a basic physical force in nature that affects all matter. This is an omnipresent force, and all living organisms on earth are subject to constant gravitational forces.^[189] Newton’s law of universal gravitation states that every point mass attracts every other point mass by a force that is proportional to the product of the masses and inversely proportional to the square of the distance between them, and is summarized by the following equation:

$$F_g=G(\frac{m_1{m}_2}{r^2})$$ (1)

where m₁ and m₂ are the masses of the objects, r is the distance between the objects, and G is the gravitational constant. The gravitational force of an object on Earth can be simply represented by F_g = mg, where m is the mass of the object, and g is the gravitational acceleration constant. Researchers have studied and manipulated this phenomenon in clever ways to engineer 3D multicellular structures.

Interestingly, numerous studies demonstrate that the nullification of gravitational effects induces the generation of 3D cellular structures. A random positioning machine (RPM), which simulates a microgravitational field in which masses experience omnidirectional gravitational force,^[190] has achieved the formation of various 3D structures.^[191-193] Presence of gravity also supports the simplest experimental protocols for engineering of 3D structures by using the natural sedimentation of cells. Primarily involving microwells, various cell assembly methods using gravitational force have been reported.^[95,194,195] Although manipulating cell culture with gravitational force allows successful implementation, to a certain extent, of cell aggregation for the development of 3D structures, it can suffer from low reproducibility owing to the relatively passive and variable procedures. Future improvements that minimize the polydispersity of the final structures may be enabled by integrating advanced systems with precise control.^[196-198]

3.2. Capillary force and surface tension

Capillary force (F_c) describes action at the liquid-air-solid interface of confined structures that arises to minimize the surface free energy of the interface. At the microscale, capillary action can become the dominant force. Surface tension relates the tendency of an interface to reduce its surface area, a critical parameter that determines capillary force. To give a theoretical perspective in terms of molecular interactions, for instance, water molecules in the bulk region experience intermolecular forces in all directions, which means that the net force is zero. However, water molecules at the interface are subject to non-zero net forces, thereby effectively pulling the interface inwards with them. The magnitude of the capillary force can be determined using the Young–Laplace equation, which is described by:

$$F_c=\frac{2\gamma \cos \theta }{r}$$ (2)

where γ is the surface tension, r is the diameter of the channel, and θ is the contact angle of the wetting liquid on the channel at the three-phase contact line.

In regard to biological microengineering, capillary force and surface tension have been used to perform cell trapping within specific regions of the microstructures. Garg et al. demonstrated capillary force-assisted seeding of adipose-derived stem cells (ADSCs).^[199] A collagen-pullulan hydrogel was placed on a wax paper, into which murine ADSCs were actively absorbed by lifting the porous hydrogel substrate. Capillary seeding achieved rapid cell entrapment (< 1 min) with a high formation efficiency (99.38%) that is significantly higher than those of other seeding approaches such as injection (~60%), centrifugal culture (~70%), and orbital culture (~80%).^[200] In a similar manner, Park et al. investigated a cell docking method using surface tension-driven capillary flow, which was performed by using the receding meniscus of a cell suspension undergoing evaporation.^[201,202] Consequently, capillary force provides an alternative route of cell trapping for facile 3D cell structure formation; it should be noted, however, that experimental conditions such as cell density and seeding flowrate need to be optimized accordingly for proper operation.

3.3. Centrifugal force

Centrifugal force (F_ω) is the force acting outward due to the inertia of a rotating object. There are not many opportunities for cells in our body to undergo centrifugal force in a normal physiological environment. Nevertheless, centrifugal force has the advantage of being able to easily produce a very large force that is hundreds to thousands of times greater than gravity.^[203] The centrifugal force that the cells experience owing to rotation is the same magnitude as the centripetal force but in the opposite direction, and can be described by:

$$\text{Centrifugal force}(F_{\omega })=\frac{m{v}^2}{r}=m{\omega }^{2}r$$ (3)

where m is the mass of the cells, v is the tangential velocity of the cell, r is the radius from the center of rotation to the cells, and ω is the angular velocity. The conventional application of such forces is in centrifuges, where substances and particles are separated.^[204] It is only recently that researchers have begun utilizing centrifugal forces in 3D cell culture systems.

Cells can sense centrifugal environments. For instance, the proliferation of HeLa cells increase by approximately 20%, 50%, and 30% in 18-, 35-, and 70-g environments, respectively, and the overall cell replication time decreases by 17% at 35 g compared to static environments. More specifically, the G1 phase duration decreases by 26%, and the S, G2, and M phases are unchanged.^[205] Another study shows an increase in proliferation of adipose-derived stem cells (ADSCs) by 13%, 26%, 44%, and 40%, respectively in 10-, 20-, 40-, and 60-g environments, respectively, compared to static environments.^[206] A more recent paper reports that an increase in length of F-actin and myosin fiber, and hence the improvement in cell motility, is observed in centrifugal conditions (< 20 g) in human microvascular endothelial cells (HMEC-1).^[207] Moreover, gene analysis showed the upregulations of genes promoting nitric oxide (NO) production, endothelial activation, and angiogenesis, which play key roles in maintaining quiescence in the vascular wall by inhibiting cell proliferation, inflammation, and thrombosis.^[208-210] Therefore, the centrifugal force environment does not have a fatal effect on cell survival, rather it increases cell proliferation and differentiation rate in some cell types.

In short, centrifugal force, which can be very large, can be an efficient method to aggregate cells efficiently. However, centrifugation may apply unwanted effects on cells, which would potentially disrupt many cellular processes. Therefore, the underlying biological mechanisms of spheroid/organoid formation, and the potential biological effects of centrifugation should be further studied.

3.4. Adhesive force

Adhesive forces influence cell-cell and cell-ECM interactions that play significant roles in many biological processes such as apoptosis, differentiation, growth, morphogenesis, wound healing, tissue preservation, and organ development.^[211-213] Adhesive forces can act at both the microscopic cellular level, and at the macroscopic cell-substrate level. In the former, cell adhesion is mediated by cell adhesion molecules (CAMs), which are transmembrane receptors involved in binding with other cells or with ECM for many essential signaling pathways. These molecules are classified into four groups: the immunoglobulin super family CAMs (IgCAMs), integrins, cadherins, and selectins.^[214,215] In addition to the different types of CAMs, cell adhesion is also dependent on fibrillar adhesions,^[216,217] and podosomes.^[218,219]

Recently, the relationship between cell adhesion and biomaterials has been extensively studied, and the results revealed that the control of adhesion depends on the properties of the biomaterials such as stiffness,^[220,221] surface charge,^[222,223] roughness,^[224,225] wettability,^[226-228] and surface modifications.^[229,230] For instance, it has been shown by using stiffness-controlled PDMS substrates that different cells respond to material stiffness differently. By blending two products such as Sylgard 184 and Sylgard 527, PDMS stiffness can be controlled from 10 to 1500 kPa; with higher stiffness PDMS substrates displayed higher level of focal adhesions and intermediate filaments of human limbal corneal epithelial cells compared with cells on higher-stiffness PDMS substrates.^[231] In addition, soft substrates can provide more cytocompatible conditions for adhesion of non-malignant epithelial cells of ureter (HCV29) and transitional cell carcinoma cells (T24), whereas higher-stiffness substrates resulted in the formation of a cell-repellent surface.^[232] Moreover, Brown et al. demonstrated that bovine vascular smooth muscle cells (VSMCs) show an increase of 39% in cell adhesion and of 42% in spreading by modulating the PDMS stiffness from 0.05 to 1.79 MPa.^[233] It was also reported that MSCs cultured on stiff substrates showed increased adhesion, which promoted differentiation into smooth muscle cells (SMC) and chondrogenic lineage.^[234] Therefore, there exists an optimal scaffold stiffness for different cells, as the stiffness of their microenvironment in vivo differ based on the specific cell types.

Many advanced surface engineering techniques have opened up opportunities for controllable cell adhesion. Different cell types of adhesion with intracellular and external components must be identified for cellular response, morphological evolution, and functional specificity. Therefore, the study of cell adhesion provides important implications for the design of appropriate substrates for tissue regeneration. In addition, various forms of cell adhesion are significant for understanding cellular mechanosensors, as well as providing the information on molecular signal pathways in cell regulation.

3.5. Hydrodynamic force

Hydrodynamic force refers to a type of force that acts on or is exerted by fluidic shear stress. This is substantially related to many essential physiological processes. The most prominent example is the cardiovascular system, where blood flow can affect mechanical signaling in epithelial and endothelial cells, such the secretion of nitric oxide, mechanical stress -associated signal transduction in arteriosclerosis, and many others.^[235-237] The fundamental basis of hydrodynamics-based microengineering platforms is to manipulate the volumetric fluid flows on a characteristic length scale in the orders of microns, such that it can apply mechanical shear force on cells in their cellular microenvironment. The fluidic shear stress (τ_s) that the cells experience can be expressed as follows:

$${\tau }_{cell}=\frac{6\mu Q}{h^{2}w}$$ (4)

where the μ is the fluid viscosity, the Q is the volumetric flow rate of fluid, and h and w are the height and width of a microchannel, respectively. Therefore, the shear stress is dependent on the microchannel dimensions and the applied flow rate, while the viscosity remains constant.

Based on this principle, mechanical shear stress can not only affect cell morphology and orientation, but also modulate cell–cell or cell–surface binding kinetics that regulate, for instance, cytoskeleton protein production, secretion, and organization.^[238] Buchanan et al. developed a microfluidic 3D co-culture system to study paracrine signaling and tumor angiogenesis using a microvascular endothelial cell line (TIME) and breast cancer cell line (MDA-MB-231), while applying varying shear stress conditions.^[239] This study showed that endothelial cells were successfully cultured in their microfluidic collagen I-based hydrogels for fluidic shear stress ranging from 1 to 10 dyn/cm². Under no-flow static conditions, only the expression of vascular endothelial growth factor A (VEGFA) was enhanced. However, in the presence of flow, many proangiogenic factors including VEGFA, matrix metalloproteinase 9 (MMP9), platelet-derived growth factor B (PDGFB), and angiopoietin 2 (ANGPT2) were all upregulated, implying that flow is a very important factor.

Moreover, shear stress influences the morphology and fate of cells. For example, Chau et al. evaluated cellular responses in HUVECs under different shear stresses ranging from 0.7 to 130 dyn/cm².^[240] HUVECs were cultured for 20-hours under perfusion, while monitoring the production of von Willebrand factor (vWF), which is a glycoprotein recognized by platelets in damaged blood vessels. The authors found that HUVECs exposed to a shear level above 5 dyn/cm² showed significantly higher vWF secretion, whereas lower levels of vWF were measured under a shear stress of 0.7 dyn/cm². In addition, the cells were 30% smaller than those in the control experiment under static conditions, which agrees with the trend of decreasing cell size with increasing shear stress. Therefore, shear stress can play a significant role not only for simulating in vivo microenvironment, but also in understanding the cellular response toward the external fluid-mechanical stimulations. One challenge is that the inherent limitation of laminar flow in rectangular microchannels results in varying fluid velocity profiles from the center to the edge, which leads to spatial difference in cellular growth and development inside the microchannels. Therefore, the development and design of round cross-section channels may be needed to mimic the microvascular model with more consistent results.^[241]

4. Conclusions and Future directions

It has been a few decades since microsystems started being applied to cellular research. Now there is an enormous number of unique microsystems that differ in terms of their engineering principles, designs, fabrication, operation, and microenvironment for conducting 3D culture of different cell types. These systems have provided versatility to the field of cellular research by providing platforms for many cell types and their biological applications. With the recent emphasis on 3D in vitro models for better disease modeling, tissue engineering, and drug screening, researchers have developed advanced microsystems that can implement complex physiological representations. Therefore, the primary aim of the review article is to provide a classification of the engineered microsystems-based 3D in vitro models in regard to their underlying physics. With this, biologists can benefit from understanding the underlying principles and get ideas on which particular microsystem would be suitable for their cells/cellular applications-of-interest. Moreover, the review can also guide engineers to assess advantages, disadvantages and similarities of such microsystems to guide future research. We categorized engineered microsystems into four types: microwell-, centrifugation-, liquid drop-, and microchannel-based culture systems, each of which offers its own advantages and disadvantages. These four systems have generated diverse types of 3D cellular structures in the forms of spheroids, multicellular spheroids, and organoids to enable novel biological studies. We also described the five mechanical principles including gravitational force, capillary force, adhesive force, centrifugal force, and hydrodynamic shear stress, each of which can uniquely influence the formation of 3D cellular structures from the microscopic cellular levels to macroscopic system levels.

Although many engineered microsystems have shown significantly promising results for generating well-defined 3D cellular structures, there remains a number of improvements to achieve more complex in vitro system. The physiological conditions such as continuous supply of nutrients, gas exchange, and vascularization are all important for closer recapitulation of in vivo environments. Recently, engineered microsystems integrated with optical tweezers,^[242,243] acoustic waves,^[244] electrodes,^[245] and 3D printed architectures^[246] have emerged as promising analytical platforms for providing deep insights on complex cellular interactions. Many of these microengineering systems, however, use PDMS as the base material, are often limited owing to its non-degradable properties.^[94,247,248] Therefore, alternative materials that are as easy to access and manipulate as PDMS should be investigated to alleviate such problems.^[249] 3D bioprinting is an effective solution to overcome some of the inherent PDMS limitations, in which precise cell layer stacking is performed using biomaterials, such as functionalized hydrogels, degradable biopolymers, and implantable cell friendly materials, resulting in the fabrication of implantable 3D building blocks,^[250,251] and artificial organs.^[252,253] Despite the promising aspects of the methods, issues regarding cost, reproducibility, and scalability all to be resolved through further research in order to fully unlock their potential.

The next step forward in the field of 3D in vitro model systems is to create a multi-organ network model to provide deeper insights into human physiology and disease mechanisms. While efficiently and reliably creating these multi-organ networks is one challenge, another issue is that researchers must ensure that the integrative system accurately reflects physiological conditions as a whole. It is very important that close collaboration among relevant researchers, such as engineers, biologists, and clinicians, is encouraged for delivering maximal benefits of such 3D in vitro model systems. For instance, engineers should interact with clinicians to provide more relevant in vitro models that meet with clinical demands. Biologists can also cooperate with engineers and clinicians to give in-depth biological knowledge in order to design more accurate in vitro models. We believe that future microengineering techniques will enable complex and sophisticated self-contained human in vitro systems for many unique applications including regenerative medicine, tissue engineering, and personalized medicine, which will be useful for the prediction of patient-specific therapeutic efficacy, and potential side effects.