Development of an Extracellular Matrix Plate for Drug Screening Using Patient-Derived Tumor Organoids

With the advances in organoid culture, patient-derived organoids are utilized in diverse fields to broaden our understanding of conventional 2-dimensional (2D) culture methods and animal models. Patient-derived organoids have found new applications, such as screening for patient-matching drugs, immune checkpoint drugs, and mutation-target drugs, in the field of drug screening. However, conventional dome-shaped Matrigel drop-based screening methods using 24-, 48-, and 96-well plates are not effective for carrying out large-scale drug screening using organoids. Here, we present a newly developed 96-well plate-based method for the effective screening of patient-derived tumor organoids embedded in Matrigel. The new screening plate has a central hole with a diameter of 3 or 5 mm to provide a definite space for placing Matrigel in a cylindrical shape. Compared to the conventional dome-shaped Matrigel where the Matrigel drop is located arbitrarily, a cylinder-shaped Matrigel position in confined central wells allowed for faster and cost-effective tumor organoid screening. Importantly, the cylinder-shaped Matrigel ensures better consistency in high-throughput image-based analysis, which is used worldwide. Our results demonstrate the possibility of replacing the conventional 24-, 48-, and 96-well plates with the newly developed plates for effective tumor organoid screening.


Introduction
Pluripotent and adult stem cell-based organ recapitulations called organoids are three-dimensional (3D) cell clusters with physiological and hierarchical cell organizations that closely mimic paired human organs [1]. There has been a remarkable progress in the field of organoids. Organoids representing various organ types have been developed through advanced studies of multi-cellular communication [2], extracellular matrix [3], genetic manipulation [4], growth conditions [5], and other factors [6][7][8][9]. In parallel to the advances in organoids, studies on proper platforms or plates for maintaining these developed organoids for high-throughput studies are essential [10]. The culture methods for organoids using conventional plates are relatively simple [11,12]. Effective culture, in terms of economy and time factors, for mass production or screening is not fully achieved. The need to filter for proper and adequately sized organoids to quantitatively assess outcomes is a burdensome and additional process in the field of tumor organoid screening [13,14]. Organoids are well known for their self-organization with Yong Hun Jung: First author. unrestricted structural, morphological, and physiological development; therefore, the diverse expansion in organoid size is natural [15,16]. However, from the perspective of organoid analysis in high-throughput screening or analysis, obtaining a uniform distribution in a single axis plane is crucial.
In most cases, 3D organoids are cultured in the extracellular matrix, in this case Matrigel, as sessile drops on a flat surface of conventional 24-, 48-, or 96-well plates, resulting in the positioning of the Matrigel in a dome shape [17,18]. This method is relatively fast, simple, and does not require any special techniques; however, a dome-shaped extracellular matrix has a limitation in high-throughput screening, because the internal gradient distribution in the organoids varies with size and is formed depending on the position of the z-axis. Simultaneous organoid culture derived from single cells in a dome shape does not ensure similar growth rates between the surface and inner parts, resulting in varied sizes and morphological differences within a single plane of a single Matrigel drop. In high-throughput analysis, where image-based analysis is used [19,20], ensuring and understanding the correct causes of growth differences are paramount for accurate screening. However, size differences between the surface and inner parts within a single Matrigel drop hinder evaluating the effects of external factors, such as drugs, on the different groups. Consequently, there is a necessity for an additional sorting process during or after experiments to quantitatively compare and analyze.
To overcome such irregularities, one possible method is to actively replenish culture components [21], considering that growth factor is not actively transferred. The factors are constantly replenished every few hours, and the organoids are cultured effectively regardless of their position. However, this is not economically effective and is not suitable for high-throughput analysis, as drug treatments have to be performed simultaneously. Once drug treatment is initiated, the reasons for size differences become ambiguous, as they could be caused either by high doses of drugs or through inefficient delivery of growth factors into the inner part of the Matrigel dome shape. The differences between the outer and inner parts could be attributed to various causes: intercellular communication [22], coffee-ring effect [23], distinct size differences of matrix pores [24], and cell density difference in the z-axis.
To acquire uniform and reliable data without any complications and disruption, we developed a screening plate called Extracellular Matrix plate (EM plate), harboring different Matrigel shapes, which is compatible with the conventional 96-well plates. It is a 96-well plate with a centered hole in each well for placing a cylinder-shaped Matrigel on a flat surface, supporting a uniform arrangement of organoids in a single plane. By altering the widely used method of sessile drop on a flat surface, we generated a high-throughput plate capable of producing an array of uniformly sized organoids, in a single plane without merging images in the z-axis. This plate would enable effective organoid culture that is costeffective and time-efficient.

Development of EM Plate
The novel EM plate is based on 96-well plates and it has two versions of a center hole. The top part of the well is shaped as an inverted pyramid to provide a definite point for changing the medium or washing out during cell analysis. To safely place Matrigel into the center hole, each well was pre-coated with 3% bovine serum albumin to prevent the Matrigel from sticking to the sides of the well. This allows for a faster and more effective seeding process because Matrigel can slide down to the center hole when roughly seeded (Fig. 1A). The EM plate was developed in two versions: the D3 EM plate with a 3-mm diametere central hole and the D5 EM plate with a 5-mm diameter central hole. Each plate serves different purposes: the D3 EM plate is suitable for screening purposes, whereas the D5 EM plate is designed for mass production of organoids (Fig. 1B, C). In the conventional plates, the dome-shaped Matrigel is placed on a flat surface in the 24-, 48-, and 96-well plates; therefore, the shape and position are arbitrary and depend on the dexterity of the user ( Fig. 1D and Fig. S1). The diameters of the Matrigel dome are dispersed over a wide range, whereas the diameters of the Matrigel cylinder were within a closed range of 3 and 5 mm (Fig. 1E). Not only were the diameters of the seeded Matrigel uniform, but the duration for Matrigel placement was also decreased significantly. The central hole of the EM plate enables easy access for the multi-pipette used for placing the Matrigel. In addition, users do not have to consider the placement of Matrigel as in the case of conventional flat-bottomed plates; therefore, the duration is greatly reduced even when manually placing organoid harbored Matrigel (Fig. 1F).

Distribution of Tumor Organoids
To demonstrate how the EM plate is better suited for highthroughput image analysis for drug screening, the distribution position of tumor organoids within the Matrigel was assessed. To ensure repeatability of the distribution position, four different dome-shaped and cylinder-shaped Matrigel entities were analyzed (Fig. 2B, D). To confirm the distribution, the dome-shaped Matrigel area was divided into five sections based on the top view in a single xy plane ( Fig. 2A). Starting with the conventional Matrigel dome shape, the average diameter of the Matrigel dome was 3 mm. The size of the tumor organoids from each section was assessed by dividing the plane into five regions. The size of organoids from the conventional Matrigel dome shape varied greatly from the outer edge to the inner center (Fig. 2C). However, the cylindrical shape of the Matrigel in the EM plate enabled a rather consistent distribution of tumor organoids ( Fig. 2D, E). Even though all organoids were grown from single cells, their sizes were different depending on the shape of the Matrigel (Fig. 2F). The size distribution is gradiently formed from edge to center; therefore, insufficient delivery of growth factors into the central part could be the reason for the smaller organoid sizes. To confirm grwoth where the position, size and morphology are all randomly distributed, scale bars: 10 mm. E The average diameter of Matrigel based on the type of plate (mean ± SEM, n = 8, p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001). F duration of seeding while using multi-piepette and a single piepette. Note that conventional 24-well plate can only utilize a single pieppete; therefore, a larger amount of time is required (mean ± SEM, n = 6, p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001)

Fig. 2
Diameter distribution of tumor organoid within the different morphological types of extracellular matrix. A Schematics on how a single plane is captured by microscope in each type of Matrigel shape. B-E average diameter of organoids within each region from each shape. Upper; dome shape where size gets smaller as region goes to center and lower; cylinder shape where size remains uniform (mean ± SEM). F bright-field images of organoids at day 0, 7, and 14 in the central regions and different points on the edges. Scale bars: 500 µm uptake of organoids from different regions, COMSOL-simulation was conducted (Fig. S2). Within boundary conditions designed same as shapes of Matrigel (Table 1), three different points were analyzed: edge, middle, and center ( Fig. S2A, C). Computational graphs showed concentration uptakes at three points were all different in Matrigel dome-shaped, whereas uptakes were all same in cylindershaped Matrigel (Fig. S2B, D). Computational graphs suggested that the rate of molecular diffusion is different based on the outer shape of Matrigel that evenly flattening surface ensures same ratio of diffusion. To confirm if diffusion can be manually enhanced, a shaker was used during organoid culture to promote active diffusion into the center of Matrigel (Fig. S3). However, no conspicuous differences were observed when compared to organoids cultured without shaking. To understand this difference further, each type of Matrigel was further divided into top, middle, and bottom based on the z-stack (Fig. 3). Three different groups were tested to see how organoid sizes differed: 24-well platebased dome shape, D5 EM plate cylinder shape, and D3 EM plate cylinder shape. The 24-well plate and the D5 EM plate accommodated 30 µl of Matrigel, whereas the D3 EM plate could accommodate 10 µl of Matrigel. Each type of wells was heighted at 0.8 mm for 24-well plate, 1.5 mm for D5 EM plate, and 1.4 mm for D3 EM plate. In case of the dome-shaped Matrigel, the sizes of organoids in the outer rim of the dome shape were generally similar. The sizes of the organoids distributed on the surface were consistent; however, the sizes of organoids varied greatly among those embedded in the central part (Fig. 3A). In case of the D5 and D3 EM plates, as the surface was evenly sustained, the distribution of organoid sizes was consistent (Fig. 3B, C). The D3 EM plate holds less Matrigel than the D5 EM plate; therefore, the average organoid size is slightly smaller; however, the consistency improves when the same numbers of cells are seeded. Regardless of the Matrigel shape, the organoid sizes increased as they approached the surface. Therefore, a cylindrical shape that offers a flat surface is better suited for high-throughput image analysis on a single plane, because it promises a balanced distribution of organoids.

Confirmation of Consistency in the Fluorescence Expression in EM Plate
One of the most widely used methods to validate the effect of drugs on target subjects is to stain drug-treated samples using LIVE/DEAD fluorescence. In case of high-throughput imaging, it is crucial to analyze the images one at a time with a definite range of planes. The dome shape does not ensure consistent quality of fluorescence images because of its size differences in a single plane. To check how the level of expression differed in each shape of the Matrigel, fluorescence images of the outer and inner parts of each shape were obtained (Fig. 4). In case of the dome shape, as the fluorescent dye cannot penetrate sufficiently into the central region, the expression level is much higher in the outer edges (Fig. 4A). In contrast, in the case of the cylindrical shape, the fluorescence dye penetrated equally, resulting in equivalent expression levels (Fig. 4B). To validate the exact difference in the expression values between the dome and cylinder shapes, the expression levels were measured considering the background expression versus target organoid expression (Fig. 4C). The variation in the expression between the edge and center in the dome shape was enormous, whereas the expression level was homogenous in the cylindrical shape.

Drug Screening of Tumor Organoids in the EM Plate
After confirming that the cylinder-shaped Matrigel promises more quantified results in high-throughput image analysis, chemical-based drug screening was carried out to determine the possibility of consistent screening (Fig. 5). Whereas conventional tumor organoid culture requires approximately 7-14 days for fully matured organoid culture, the tumor organoids cultured for drug screenings were cultured for three days before drug treatments to distinguish better drug responses at day 9 ( Fig. 5A) [13,27,28]. The bright-field images clearly show how the organoids gradually undergo cell death and constrained growth based on the drug density gradient (Fig. 5B). To ensure that the results of drug screening on the EM plate were reproducible, biological replicates were evaluated (Fig. 5C). In the detailed biological replicate graphs where actual IC 50 values are marked (Fig. 5D), the first and second screening trials showed no distinct differences in the IC 50 values.

Discussion
Patient-derived organ-mimicking multi-cellular clusters called organoids are closely studied as crucial preclinical test models for drug discovery and screening. High-throughput screening of organoids is a precondition for drug selection for in vivo PDX models or patient trials. However, current organoid culture systems that rely on embedding Matrigel as a dome shape are not suitable for high-throughput analysis based on images taken, efficiency, user-friendliness, and consistency, because of heterogeneity in tumor formation. Improvements in culture plates based on conventional products for the needs of better organoid screening are lacking. The profound morphological, size-related variations, as well as user-dependent dexterity in the existing culture systems need to be addressed. In this study, we developed a 96-well plate compatible EM plate to enable relatively uniform culture of tumor organoids. The formation of cylinder-shaped Matrigel in the EM plate was achieved using a confined well located in the middle of the 96-well plate. In addition, an inverted pyramid structure on top of the center hole allows a definite point to exchange the culture medium at the junction where an inverted pyramid structure meets the vertical wall. By developing the EM plate, we are able to meet the needs for improvements in the conventional method. First, as EM plates provide definite positions for Matrigel to be placed in in dome-shaped Matrigel could be attributed to several reasons, such as insufficient delivery of growth factors, the height difference between the center and edge in the z-axis, and the coffee-ring effect. The organoids generated in the cylinder shape within the EM plate were morphologically and functionally comparable to the organoids generated in the dome shape, but the level of fluorescence was different between the two models; the cylinder shape maintained a homogenous expression throughout a single plane. The reasons for these differences in expression levels and sizes remain unclear, because various factors and intercellular components are closely related. Hypothetically, the primary factor could be the growth factor gradient within the Matrigel volume. Matrigel is an extracellular matrix; therefore, its internal structure is formed of minute fibers that closely overlap with each other to form countless pores. These pores are stacked at a higher density depending on the z-axis; therefore, growth factor diffusion might be hindered. To overcome such insufficient factor delivery, culture medium could be replenished more often [21]. However, such measures require users or systems to replenish the culture medium within a few hours, which is unsuitable for high-throughput screening. Furthermore, as numerous organoid types have been developed, the factors needed for robust growth vary tremendously, and replenishing them in a short time may result in huge expenses. In addition to this factor, there could be many other factors that affect the growth difference in dome shapes. Our results directly indicate the geometrical and morphological differences between the dome and cylinder shapes of Matrigel.
To further understand the underlying mechanisms, additional studies are needed to verify the influence of Matrigel shape on cell distribution, cell-cell interaction, and factor transfer diffusiveness. Most importantly, systematic comprehensive experiments are required to unveil the complicated features of cellular and molecular communication within the extracellular matrix.
In summary, our study introduces a new parameter for reliable high-throughput screening using organoids. Our study offers insights into how the shape of Matrigel affects the organoid distribution within the matrix and hinders faster and more reliable drug screening, using tumor organoids. The physically altered matrix shape supports the demand for high-throughput analysis using organoids. The existing method is composed of redundant steps that leads to inefficient work. In future, the development of organoid-related plates will lead to the development of comprehensive culture platforms capable of recapitulating more clinically relevant high-throughput models.

Human Tissues
All samples used for organoid establishment and biological analysis were collected from Yonsei University Severance Hospital (IRB # 0000-0003-0180-8565). The ethical review boards of both institutes approved this research. Endoscopic biopsies and surgical resections were performed to obtain samples of healthy and malignant colonic tissues.

Organoid Culture
Fresh tumor tissue samples were handled as described previously, with few modifications. The tissue was cut into small pieces, rinsed with ice-cold PBS, and digested with the gentleMACsTM kit for 60 min using the gentleMACsTM Dissociator (130-093-235, Miltenyi Biotec). The supernatant was collected, filtered through a 100 µm strainer (352360, Corning), and centrifuged at 200×g for 3 min at 4 °C. The cell pellet was suspended in Matrigel (growth factor decreased; Corning 356231) and distributed onto two 24-well culture plates at a concentration of 200,000 cells per 35 µL of Matrigel. After Matrigel polymerization, the cells were coated with culture fluid and cultured in 5% CO 2 , and the medium was replaced every 2 or 3 days. The organoids were cultured for three passages in media that fulfilled the minimal niche requirements before being used for further analysis under minimal essential conditions.

Extracellular Matrix Plate Preparation and Organoid Culture
The EM plate was manufactured by Microfit (Hanam, Korea). The plate was first treated with 100% alcohol to prewash any dust or debris within the wells. The plate was incubated overnight at 37 °C, followed by three repetitive washes using distilled water. Finally, enzymatically dissociated single cells were seeded at a density of 500-2000 cells per well in either 10 µl of Matrigel for the 3 mm EM plate and in 30 µl of Matrigel for the 5 mm EM plate. After 10 min of gelation, the complete culture medium was added for further culturing.

Drug Screening
Organoids were extracted from Matrigel and dissociated into single cells using TrypLE (together with ROCKi) and mechanical dispersion. The cell suspension was resuspended in full media. Each well of the 3 mm EM plate was seeded with 1000 cells and the cells were allowed to develop for three days. Every three days, a dilution series of doxifluridine (S2045, Selleckchem) or vehicle (DMSO) was applied to the organoid cultures. To determine the pharmacological response of organoids, cell viability was measured 6 days after treatment using the CellTiter-Glo assay (Promega) and live/dead staining (L3224, Invitrogen), according to the manufacturer's instructions. GraphPad Prism was used for both data analysis and IC 50 value estimation (version 8.0.1).

Viability Analysis (Live/Dead Staining and Cell-Titer-Glow Assay)
For LIVE/DEAD labeling, organoids were incubated at 37 °C for 40 min with 50 mM calcein-AM and 25 mg/ml ethidium homodimer-1 (EthD-1; Molecular Probes, USA) and then photographed using confocal microscopy (Olympus, Japan). ImageJ software (NIH, Bethesda, MD, USA) was used to quantify calcein-AM (green, for live cells) and EthD-1 (red, for dead cells) signals. In addition, the vitality of the organoids was quantified using the CellTiter-Glo assay (Promega). Before measuring luminescence with a microplate reader (Thermo Fisher, USA), the plates were agitated at room temperature for 30 min. A dose-response curve was generated using GraphPad Prism to determine IC 50 values.

Expression intensity calculation
Images from LIVE/DEAD assay are selected through Image J software (NIH, Bethesda, MD, USA). From the Analyze menu select "set measurements", area integrated intensity and mean gray value are selected. After selecting regions with cells, background regions are also selected from region next to cells. The expression intensity of LIVE/DEAD can be calculated by this equation: Integrated density − (area of selected cell × mean fluorescence of background readings).

Organoid-Distribution Evaluation
Single cells dissociated from organoids were embedded in Matrigel at a seeding density of 2000 cells per well. The cells were then grown in a humidified CO 2 incubator at 37 °C, with medium changes every 3-4 days. Fourteen days after plating, the distribution of the organoids in each well was examined. Microscopic (Aliened Genetics, Korea) images of the organoids from each well were acquired and the images were processed using Image J software (NIH, Bethesda, MD, USA) [26].

Statistical Analysis
Statistical analyses were performed using Prism 6.0 software (GraphPad Software Inc., San Diego, CA, USA). The data are represented as mean ± standard deviation (SD) or mean ± standard error of the mean (SEM) with at least three independent replicates. Multi-group analyses were conducted using a one-way analysis of variance (ANOVA) followed by a Student's t test. p values less than 0.05 were considered statistically significant *p 0.05, **p 0.01, and ***p 0.001.