Recent advances in microarray 3D bioprinting for high-throughput spheroid and tissue culture and analysis

Abstract

Three-dimensional (3D) cell culture in vitro has proven to be more physiologically relevant than two-dimensional (2D) culture of cell monolayers, thus more predictive in assessing efficacy and toxicity of compounds. There have been several 3D cell culture techniques developed, which include spheroid and multicellular tissue cultures. Cell spheroids have been generated from single or multiple cell types cultured in ultralow attachment (ULA) well plates and hanging droplet plates. In general, cell spheroids are formed in a relatively short period of culture, in the absence of extracellular matrices (ECMs), via gravity-driven self-aggregation, thus having limited ability to self-organization in layered structure. On the other hand, multicellular tissue cultures including miniature tissues derived from pluripotent stem cells and adult stem cells (a.k.a. ‘organoids’) and 3D bioprinted tissue constructs require biomimetic hydrogels or ECMs and show highly ordered structure due to spontaneous self-organization of cells during differentiation and maturation processes. In this short review article, we summarize traditional methods of spheroid and multicellular tissue cultures as well as their technical challenges, and introduce how droplet-based, miniature 3D bioprinting (‘microarray 3D bioprinting’) can be used to improve assay throughput and reproducibility for high-throughput, predictive screening of compounds. Several platforms including a micropillar chip and a 384-pillar plate developed to facilitate miniature spheroid and tissue cultures via microarray 3D bioprinting are introduced. We excluded microphysiological systems (MPSs) in this article although they are important tissue models to simulate multiorgan interactions.

Spheroid culture techniques for high-throughput screening of compound libraries

Due to the lack of physiological similarity to in vivo, the paradigm has been changed from two-dimensional (2D) cell cultures to three-dimensional (3D) cell cultures, such as cell spheroids, organoids, and bioprinted tissue constructs. There is growing evidence that 3D cell culture models are a better mimic of tissue structure and function in vivo [1–4]. Thus, 3D cell models have become extremely relevant from basic research to drug testing/screening and translational purposes. In this introductory section, we summarize culture platforms commonly used for cell spheroids and explain their pros and cons for high-throughput screening (HTS) applications.

Several scaffold-free platforms have been widely used to generate spheroids, which include ultralow attachment (ULA) well plates [5], hanging droplet plates [6], spinner flasks [7], rotating wall vessels [8], and magnetic levitation [9] (Figure 1). For example, the suspension of cells can be added in ULA 384-well plates and incubated for several days to form cell spheroids by gravity [10]. The formation of cell spheroids can be accelerated by generating cell pellets by centrifugation of cell suspension in AggreWell plates [11]. This method is relatively simple and straightforward for changing growth media and requires less resources, thus widely used for HTS of compounds against cell spheroids. In hanging droplet plates, cells can be settled down at the round bottom of water droplets by gravity and form spheroids [12]. Due to difficulty in changing growth media and instability of hanging droplets, this method is used mostly for short-term culture of cell spheroids. Spinner flasks and rotating wall vessels are used for large-scale culture of cell spheroids and unsuitable for HTS of compounds. Cell spheroids are formed by applying convectional force with continuous stirring [12] or constant circular rotation [8]. In magnetic levitation, spheroids are generated by mixing cells with magnetic nanoparticles and applying external magnetic force for aggregate formation [13]. Although it may reduce necrosis in the core of large cell spheroids, cell viability and imaging could be interfered with nanoparticles, thus having limited utility [14]. Finally, spheroids can be formed by culturing cells in hydrogels such as alginate, collagen, fibrin, and hyaluronic acid [15]. Cells suspended in biomimetic hydrogels can be dispensed in 24-/48-well plates and cultured for 1 or 2 weeks to form spheroids. Although this method could provide biomimetic environments of extracellular matrices (ECMs) in vitro by using hydrogels and growth factors, it is cumbersome to dispense viscous hydrogels accurately and change growth media without disturbing cells in hydrogels, thus not easily amenable to HTS.

Figure 1. Conventional scaffold-free spheroid culture methods.

(A) Hanging drop plate, (B) spinner flask and rotating wall vessel, (C) ULA well plate, and (D) magnetic levitation.

Advanced 3D cell/tissue models and their technical challenges for HTS

Although spheroid models are simple, highly reproducible, and inexpensive for HTS assays, they have limited in vivo-like tissue structure and function (Table 1). Thus, more advanced 3D cell/tissue models have been developed for in vitro disease modeling and predictive compound screening. In this section, we introduce human organoids and 3D bioprinted tissue constructs and explain their technical challenges for HTS assays.

Table 1.

Advantages and disadvantages of spheroid and organoid models

Cell model	Advantages	Disadvantages
Spheroid	Generated from a wide range of cell types by cell–cell adhesion with one or multiple cell types Simple and small 3D cell models with high reproducibility Inexpensive as compared with organoid models Compatible with HTS of compounds	Simple cell aggregates, thus having limited ability of self-organization Limited in vivo-like tissue structure, polarity, and function Difficult to mimic cell–ECM interactions due to the formation of spheroids in scaffold-free conditions
Organoid	Generated from pluripotent stem cells, adult stem cells, and progenitor cells by following developmental processes Complex tissue models with various organ-specific cell types and ECMs Recapitulating in vivo-like tissue structure, polarity, and function by self-organization Inexpensive as compared with animal models Suitable for studying developmental biology and morphogenetic events at the organ/tissue level Suitable for studying patient-specific disease mechanisms and personalized cancer treatment	Still immature as compared with tissues in vivo with vascular structure Poor reproducibility and scalability due to long-term cell culture, thus requiring culture standardization Cell death in the core of organoids due to diffusion limitation, thus requiring dynamic cell culture and organoid size control Low-throughput organoid imaging and analysis for quality control Incompatible with HTS yet

Organoid culture

Organoids refer to miniature tissues differentiated from pluripotent stem cells and adult stem cells into brain, lung, liver, stomach, intestine, kidney, and pancreas [16,17]. Human organoids contain multiple organ-specific cell types and have the cellular architecture that is strikingly similar to the human organ in vivo. While human organoids represent a new and innovative direction in predictive screening of drug candidates, there are several technical challenges to adopt human organoids in the drug discovery process [18]. Current organoid cultures rely on the ability of stem cells to self-organize into discrete tissue structures with a step-wise process that mimics normal organ development [19]. Owing to the long-term culture of organoids (typically 3 weeks–3 months) in biomimetic hydrogels (e.g., Matrigel) to enhance organoid maturity, there are batch-to-batch variations in function, which present major obstacles for quantitative analysis and reproducibility of results. Due to the diffusion limitation of oxygen and nutrients into the core of organoids [20,21], it is necessary to control the size to obtain reproducible organoids. In addition, the methods of analyzing organoids are still low throughput, which is a major obstacle to conduct rapid drug screening in vitro.Currently, structural and cellular characterization of organoids are performed by labor-intensive cryosectioning followed by immunostaining, which need to be replaced with high-throughput, whole organoid staining and imaging [22]. Current organoid cultures in 6-/24-well plates, Petri dishes, and spinner flasks are highly labor-intensive and unaffordable for HTS due to a large volume of expensive cell culture media and growth factor cocktails required. There are still lack of miniature and high-throughput culture systems that allow small-scale organoid culture for HTS and multiple organoid interactions for modeling complex diseases. Thus, there is a great potential that microarray 3D bioprinting could address the technical challenges in organoid culture.

3D bioprinting techniques for fabricating tissue constructs

Since the concept of 3D bioprinting was defined in the first bioprinting international conference at the University of Manchester, U.K. in September 2004, there have been several 3D bioprinting techniques developed to facilitate cell spheroid and tissue cultures rapidly and robustly for tissue engineering and regenerative medicine, cancer research, high-throughput drug screening, and transplantation and clinics [23,24]. Stem cells, progenitor cells, primary cells, and cancer cell lines suspended in hydrogels have been dispensed on glass slides, on agar plates, in 6-/24-well plates, and on pillar plates by contact and non-contact printing methods [25,26]. Among the cell printing methods (Figure 2), extrusion-based 3D bioprinting has been widely used as it can generate relatively large tissue constructs quickly by contacting the extrusion nozzle to the receiving surface and dispensing cells in hydrogels layer-by-layer. The extrusion-based 3D bioprinting technique allows to generate thick tissue constructs in various geometries rapidly and can handle highly viscous hydrogels at high cell density easily. However, the viability of extrusion-printed cells is relatively low (40–80%) due to high shear stress in the nozzle and toxic radicals generated from photo-initiators in the photopolymerization process commonly used to encapsulate cells in hydrogels [27]. There might be cell death in the core of relatively large, bioprinted tissues due to diffusion limitation of oxygen and nutrients. Among non-contact printing methods, piezoelectric nozzle-based bioprinting, solenoid valve-based bioprinting, and laser-induced forward transfer (LIFT) bioprinting are the most commonly used. Hydrogel droplets containing cells are ejected and deposited to the receiving surface without direct contact, which lead to the formation of 3D structure after hydrogel gelation. Since the volume of hydrogel droplets from piezoelectric nozzle-based bioprinting could be as small as 100 picoliter, it can generate high resolution of small tissue constructs at high cell viability (>90%). However, it is difficult to print high density of cells in viscous hydrogels due to frequent cell precipitation in the cartridge reservoir and nozzle clogging. In addition, the biofabrication speed of piezoelectric printing is much slower than that of extrusion printing due to extremely small droplets dispensed [28]. Solenoid valve-based bioprinting is superior to piezoelectric nozzle-based bioprinting in terms of cell printing reproducibility. Since the volume of hydrogel droplets from solenoid valve-based bioprinting could be as small as 20 nanoliter and as big as 5 microliter, it can generate small and big tissue constructs relatively fast and maintain high cell viability after printing due to relatively low shear stress within the solenoid valve. However, it is somewhat difficult to print high density of cells in viscous hydrogels due to valve clogging. The LIFT bioprinting technique is similar to thermal inkjet printing, which works by local heating and bubble formation. In LIFT, hydrogel droplets containing cells are ejected by focusing a pulsed laser beam on an absorbing quartz disk coated with a hydrogel containing cells, generating a bubble by local heating, exploding the bubble, and transferring the hydrogel in the bubble to the receiving surface. LIFT bioprinting requires delicate control of laser intensity, moisture, viscosity of the hydrogel, and thickness of the hydrogel layer to print hydrogel droplets accurately and maintain high cell viability [29].

Figure 2. Four representative 3D bioprinting techniques developed to fabricate tissue constructs.

(A) Extrusion-based, (B) piezoelectric nozzle-based, (C) solenoid valve-based, and (D) laser-based.

In general, 3D bioprinting technology allows to produce engineered tissue constructs reproducibly by accurately positioning different types of cells in hydrogels layer-by-layer. Bioprinted tissue constructs with high cell density can be cultured in vitro or implanted in animals for tissue maturation. A wide variety of tissues have been successfully printed to recapitulate the structural, functional, and mechanical integrity of complex organs such as liver, skin, cartilage, blood vessels [30,31]. For example, Organovo marketed ExVive™ human liver tissues in a 24-well plate generated by 3D bioprinting for better mimicking the complexity of the liver tissue structure in vivo and providing high predictivity of human liver toxicity [32]. However, current 3D bioprinting approaches are poorly suitable for HTS assays yet due to several technical challenges. In general, HTS assays require cells dispensed into traditional 96-/384-/1536-well plates, which should be imaged by automated fluorescence microscopes and measured by microtiter well plate readers to obtain cell images and signals rapidly. Bioprinted tissue constructs are often too big in size (typically 1 cm3 or bigger) to be printed and cultured in 96-/384-/1536-well plates and require large amounts of expensive human cells for HTS assays. Therefore, scale-up production of bioprinted tissue constructs for HTS assays is still challenging. In addition, it is difficult to achieve high reproducibility necessary for HTS assays (i.e., coefficient of variation or CV < 25%) from bioprinted tissue constructs due to long differentiation and maturation processes necessary. Furthermore, monitoring cells and spheroids in bioprinted tissue constructs in real time during cell growth and differentiation pose significant challenges because 3D cells are not grown in a single focal plane and their size is too big and opaque. Bioprinted tissue constructs are not amenable to currently available HTS imaging systems. To alleviate these technical challenges, miniature tissue constructs (as small as 1 mm3) have been generated mostly by droplet-based 3D bioprinting such as solenoid valves and piezoelectric nozzles (‘microarray 3D bioprinting’) [33]. The microarray 3D bioprinting technology has been demonstrated by dispensing single or multiple layers of cell types in biomimetic hydrogels on microtiter well plates and pillar plates and applied for predictive compound screening. In the following section, we introduce how microarray 3D bioprinting can be used to improve assay throughput and reproducibility for high-throughput, predictive screening of compounds.

Droplet-based, miniature 3D bioprinting (‘microarray 3D bioprinting’) and its applications

Droplet-based bioprinting has been initially demonstrated for dispensing and immobilizing microarrays of DNAs [34,35], RNAs [36], carbohydrates [37,38], antibodies [39,40], ECMs [41], and enzymes [33,42] on functionalized glass slides and silicon chips. By multiplex fluorescent detection methods, molecular levels of interactions have been investigated on microarrays, producing an enormous amount of information simultaneously. In 2008, Lee et al. [43] demonstrated rapid cell printing and encapsulation in biomimetic hydrogel spots on functionalized glass slides (‘cell microarrays’) by using solenoid valve-based, miniature, droplet bioprinting (‘microarray 3D bioprinting’). Bioprinted breast cancer cells in collagen were formed spheroids after several days of culture in growth media and coupled with enzyme microarrays for metabolism-induced toxicity assays. This microarray 3D bioprinting technology could offer several advantages over more conventional extrusion-based 3D bioprinting [44]. Specifically, it requires small amounts of cells, growth media, compounds, and reagents for cell culture and analysis. As it can seed and produce a uniform size of spheroids or organoids in high-density microtiter well plates or 384-pillar plates efficiently, it is well-suited for HTS of compound libraries, which include drug candidates, chemicals, cosmetic ingredients, and environmental toxicants. To scale-up, the generation of spheroids and organoids in HTS systems reproducibly, microarray 3D bioprinting technology has gained tremendous interest in the field of biomedical sciences [26]. In addition, the cell encapsulation technology is flexible with various hydrogels, including alginate, collagen, Matrigel, fibrin etc. and allows for culturing multiple cell types from different organs in hydrogel droplets into spheroids, thus providing more predictive insight into potential organ-specific toxicity of compounds. Miniature 3D cell cultures may also simulate an ECM environment in vivo by using biomimetic hydrogels and therefore help to maintain the specific biochemical functions and morphological features of human cells similar to those found in human organ tissues. High-throughput, high-content 3D cell imaging could decipher toxicodynamic and toxicokinetic traits of compounds and help us to understand complicated toxicology pathways and related adverse responses in early stages of research.

Since then, there have been several studies demonstrated on high-throughput culture of spheroids and organoids on Petri dishes, glass slides, microtiter well plates, micropillar/microwell chips, and 384-pillar plates by microarray 3D bioprinting technology (Figure 3). We summarize here how microarray 3D bioprinting along with high-throughput image analysis has been performed on different platforms for spheroid and organoid culture and analysis for HTS of compounds.

Figure 3. Several platforms used for microarray 3D bioprinting for high-throughput spheroid and organoid culture.

(A) Petri dishes for direct printing of cell-laden hydrogels, (B) 24-/48-well plates for direct cell printing in hydrogels, (C) micropillar chip and microwell chip for miniature spheroid culture, and (D) 384-pillar plate complementary with standard 384-well plates for high-throughput spheroid culture and analysis.

Petri dishes

In 2011, Xu et al. [45] printed mouse embryonic stem cells (mESCs) suspended in media on a petri dish lid by microvalve-based bioprinting, inverted the lid, and created hanging droplets with the cells for the formation of uniform cell aggregates by gravity. The aggregates formed were transferred to ULA 96-well plates and cultured to generate embryoid bodies. It is a unique method of creating spheroids on a Petri dish lid without using hydrogels or hanging droplet plates. In 2012, Rodŕıguez-Dévora et al. [46] demonstrated microarray printing of bacterial cell suspension on glass slides in a Petri dish using a thermal inkjet printer. They printed three layers on the same spot with the first layer of suspension of Escherichia coli in Trypticase™ soy agar, the second layer of 0.3% alginate, and the third layer of antibiotics with 1.4% calcium chloride for HTS of compounds. In 2015, Ling et al. [47] generated spheroids of MCF-7 human breast cancer cells by printing arrays of cell-laden 3% gelatin solution on to a Petri dish by solenoid valve-based bioprinting.

Microtiter well plates

In 2015, Tseng et al. [48] demonstrated a unique method of generating spheroids in ULA 96-well plates rapidly by using ‘magnetic 3D bioprinting’ technology. Murine embryonic fibroblasts were magnetized using nanoparticles, printed into ULA 6-well plates, levitated for 1 h by using a magnetic drive atop, and then transferred to ULA 96-well plates by placing the plate atop a magnetic drive to attract cells to the bottom to form spheroids. In 2018, Hou et al. [49] implemented a similar magnetic drive method for consistent production of pancreatic tumor organoid models in ULA 384- and 1536-well plates. Suspension of nanoparticle magnetized cells were printed into the ULA plates, followed by using a magnetic drive to allow cells to form aggregates. This magnetic 3D bioprinting technology was further improved for large-scale production of spheroids to execute HTS [50]. In 2020, Utama et al. [51] demonstrated embedded tumor spheroid culture by printing calcium chloride and alginate in a 96-well plate, forming alginate cups by gelation, and printing high-density cell suspension in the center of the alginate cups by solenoid valve-based bioprinting for cell aggregate formation by gravity. In 2018, Czerniecki et al. [52] generated kidney organoids by printing human pluripotent stem cells into 96- and 384-well plates coated with GelTrex® using a liquid-handling robot and performed HTS of compounds. Optimization of differentiation conditions and dose-dependent compound toxicity have been performed with high-content imaging of organoids. This simple method could be widely applicable if organoids can be derived from stem cells without encapsulation in biomimetic hydrogels. In 2020, Renner et al. [53] successfully demonstrated homogeneous generation and maintenance of human midbrain organoids in a conical 96-well plate by printing small molecule neural precursor cells (smNPCs) with 0.4% polyvinyl alcohol using an automated liquid handling system to increase cell–cell adhesion and aggregation. In addition, they also generated cortical organoids in ULA U-bottom plates from smNPCs by implementing the same automated workflow. It is worth noting that traditional liquid handling systems are designed for dispensing cell suspension in growth media in microtiter well plates. Thus, it would be challenging to print cells in hydrogels layer-by-layer for generating miniature tissues. However, it can be a powerful tool to dispense cell suspension in ULA well plates or biopolymer-coated well plates and create spheroids for HTS of compounds.

Micropillar/microwell chip

In 2014, Kwon et al. [54] developed a micropillar/microwell chip platform that couples microarrays of recombinant adenoviruses carrying genes for several cytochrome P450 isoforms (CYP450s) with human liver cell microarrays, to assess differential drug metabolism. THLE-2 cells that are closely related to human hepatocytes, yet have insignificant levels of Phase I and II drug metabolizing enzyme (DME) activities, were mixed with Matrigel and printed on the micropillar chip by solenoid valve-based 3D bioprinting for gene expression study. The controlled expression of individual and multiple reporter proteins (GFP and RFP) and CYP450 enzymes (CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP3A4, and CYP2E1) has been demonstrated in THLE-2 cells on the pillar plate by altering the multiplicity of infection (MOI) of the various recombinant adenoviruses in the microwell chip. The microarrays of 3D human cells expressing DMEs were tested with model compounds to simulate enzyme-specific hepatotoxicity. The present study demonstrated that the new chip platform can provide critical information necessary for evaluating metabolism-mediated toxicity in a high-throughput manner.

In 2018, Joshi et al. [55] demonstrated spheroid culture of neural stem cells on the micropillar chip by printing ReN-cell VM in alginate and a mixture of alginate and Matrigel using solenoid valve-based bioprinting. Alginate printed formed a gel on a poly(maleic anhydride alt-1-octadecene)-coated micropillar chip by calcium chloride and attached robustly by ionic interactions with poly-l-lysine added. Similarly, spheroids of Hep3B human hepatoma cells on the micropillar chip were generated by printing Hep3B cells in a mixture of alginate and fibrinogen via solenoid valve-based bioprinting for high-content imaging assays [56]. In 2018, Yu et al. [57] demonstrated layer-by-layer printing of different human cell types into the microwell chip by solenoid valve-based bioprinting in order to improve 3D image acquisition through deconvolution algorithm.

The 384-pillar plate

In 2018, Yu et al. [58] demonstrated spheroid culture of human embryonic kidney (HEK) 293 cells by printing the cells in a mixture of alginate and Matrigel on a 384-pillar plate using solenoid valve-based bioprinting. The 384-pillar plate with HEK 293 cells was coupled with a 384-well plate containing model compounds and five representative Phase I and II DMEs to assess metabolism-induced toxicity. Since the 384-pillar plate was built on the footprint of conventional 384-well plates, it was compatible with HTS of compounds to better predict the toxicity of parent compounds and their metabolites in vivo. Similarly, in 2020, Joshi et al. [59] demonstrated spheroid culture of human neural stem cells on the 384-pillar plate by printing ReNcell VM in a mixture of alginate and GelTrex and performed high-content imaging assays for investigating mechanisms of compound neurotoxicity.

Future directions

Recently, there have been significant advances made in in vitro disease models, including bioprinted tissue constructs with cells obtained from patients, human organoids, and multilayered cells in microphysiological systems (MPSs). These new and innovative technologies, however, still lack reproducibility, tissue maturity, enough throughput, and user-friendliness to enable rapid identification of high-quality therapeutic candidates, particularly when a disease involves multiple organ interactions. To address these technical challenges, there have been multiple attempts made to combine two or three of the technologies and enhance tissue complexity for predictive disease modeling. For example, MPSs have been used recently to culture kidney and brain organoids and enhance tissue maturity [60,61]. However, their use for organoid culture is still limited due to low throughput and user unfriendliness of MPSs. Labor-intensive, manual stem cell, and organoid loading into MPSs is a big challenge to facilitate HTS of compounds. In addition, our group has designed and manufactured novel pillar and perfusion well plates recently via plastic injection molding and demonstrated microarray 3D bioprinting of stem cells and organoid culture in static and dynamic culture conditions [62]. New multiorganoid culture systems could simulate the interactions between different organs, allowing predictive assessment of drug effects when administered in the human body and could fill the gap between in vitro cell-based tests and in vivo animal tests [63]. We envision that microarray 3D bioprinting could play a significant role in scale-up production of miniature human tissues and enhance throughput and reproducibility of data for 3D cell-based HTS of compounds.

Summary.

• There is an urgent need for physiologically relevant 3D cell culture in vitro for predictive assessment of drug efficacy and toxicity.

• Droplet-based, miniature 3D bioprinting (‘microarray 3D bioprinting’) could be used for scale-up production of miniature human tissues and improve assay throughput and reproducibility for high-throughput, predictive screening of compounds.

• To further improve tissue maturity and simulate organ interactions, there is a new trend in combining 3D bioprinting, organoid culture, and dynamic culture in MPSs.

Abbreviations

DME

drug metabolizing enzyme

ECM

extracellular matrix

HEK

human embryonic kidney

HTS

high-throughput screening

LIFT

laser-induced forward transfer

MPS

microphysiological system

smNPC

small molecule neural precursor cell

ULA

ultralow attachment

3D

three-dimensional