Download PDF

Angeline Lim, Jason Baade, Aditya Katiyar, Prathyushakrishna Macha, Zhisong Tong, Oksana Sirenko | Molecular Devices Elizabeth Fraser | Cellesce

Introduction

Patient derived organoids (PDOs) represent a promising tool to reduce pipeline attrition in drug discovery. These tumor organoids are multicellular mini replicas of the 3D tumor and has been shown to retain its in vivo characteristics.1 Studies show that patients and their derived organoids respond similarly to drugs. PDOs are therefore an advanced and biologically relevant, in vitro model for the prediction of therapeutic efficacy and toxicity. However, challenges such as assay reproducibility, scalability, and cost have limited the use of PDOs in mainstream drug discovery pipelines.

To address the challenge associated with scalability, Cellesce developed a unique industrial bioprocess for the large-scale expansion of PDOs. The organoids are grown in an environment that ensures constant delivery of nutrients and growth factors while preventing the accumulation of toxins in the culture, which can lead to cell death. This proprietary method generates highly standardized assay ready PDOs at scale, enabling applications such as high-throughput screening for drug discovery.

To demonstrate the utility of these PDOs in highthroughput applications, colorectal cancer (CRC) PDOs were seeded in multi-well plates with both manual (384 well plate) and automated approach. For automation, the pipetting function of the BAB400 bioprinter was used to dispense Matrigel domes mixed with organoids in the center of each well of a 96-well plate. Organoids were treated with selected anti-cancer drugs at various concentrations. The PDOs were monitored over time using transmitted light imaging. A deep learning-based image segmentation model was developed and used for the analysis of the PDOs. Measurements such as size, texture, intensity and other morphological and phenotypic readouts were obtained. A viability assay was carried out using live/ dead cell dyes and the PDOs were imaged in 3D on a high content confocal imager.

Figure 1. Patient derived CRC organoid (Cellesce) are supplied in cryopreserved vials (100,000 organoids). Information such as mutational profiles, tumor characteristics and morphology are available for each line.

Figure 2. Workflow for using assay ready colorectal cancer organoids.

Materials and methods

Cell culture

Colorectal cancer organoids (ISO68 line, Cellesce) were handled according manufacturer’s instructions (Figure 1). Briefly, organoids were thawed quickly at 37°C, gently resuspended and washed in media. Pellet containing organoids were resuspended in Matrigel and then seeded in 384 well plate, at 200 organoids per well. Organoids were incubated with media containing ROCK inhibitor for 48hours to improve recovery. Organoids were then treated with selected compounds for 5 days, at varying concentrations and in quadruplicates (Figure 2).

For automated seeding, the organoid Matrigel suspension was seeded into 96-well plates using BAB400 (Advanced Solution) pipette tool. Gripper (PnP tool) Sequence can be used to de-lid and lid the plates, and to move the plates to the integrated ImageXpress® Micro Confocal High-Content Imaging System for imaging. 7µl of organoid suspension was dispensed in the middle of each well to form “domes”. The automation path mapping the tool coordinates was set and recorded before the run.

High throughput Imaging and analysis

The effects of compound treatment was monitored over time using the ImageXpress® Micro Confocal system. CRC organoids were imaged using 4X objective, with z-stacks enabled. For viability assay, organoids were incubated with Hoechst, Calcein AM and ethidium homodimer for 2hrs at 37°C. Images were acquired at 10X with Z-stacking. For staining with phalloidin, organoids were fixed in 4% PFA and then incubated with phalloidin488. Images were then acquired at 10X or 20X (with water immersion objectives).

The IN Carta® Image Analysis Software was used to analyze images acquired during monitoring. A deep learning-based approach was used to create a model for organoids segmentation.

Results

Assay set up with colorectal cancer organoids

To evaluate the use of assay ready cancer organoids, we designed a proof-of-concept study in a 384 microwell plate assay. Organoids were thawed, mixed with Matrigel and seeded. After 48 hours, CRC organoids were treated with a selection of nine compounds at 7 concentrations with four technical replicates (Figure 3). After 5 days, a live/dead assay was carried out to determine the effects of the various compounds on organoid viability (Figure 5). Following the viability assay, organoids were fixed and then stained with phalloidin to observe compound effects on organoid morphology (Figure 6).

Figure 3. Assay setup. Plate map view of organoids in 384W plate. Inset shows example image (top: fluorescent, bottom: transmitted light) from one of the wells. Organoids were labelled with Hoechst (nuclei, blue), phalloidin 488 (actin, green) and ethidium homodimer (dead cell marker, red).

Monitoring of phenotypic effects of compounds on organoids using deep learning

Because PDOs do not express any fluorescent markers, brightfield imaging was used to monitor the growth of PDOs over time (Figure 4). To monitor the quality of developing organoids, we used a deep learning-based segmentation approach to analyze the acquired images. Growth of CRC organoids can be monitored by measuring their diameters or areas over time. The effects of compounds on organoids can also be quantified. Here, the growth of CRC organoids was inhibited by romidepsin and trametinib.

Figure 4. A) Example images of CRC organoids in Matrigel. Organoids were treated with the indicated compounds and monitored over 5 days. B) Overview of the SINAP workflow in IN Carta software to generate a model for organoid segmentation. C) Images acquired in transmitted light usually have high, non-homogenous background, edge effects and artifacts (such as bubbles) which prevents robust object segmentation. Shown here are example organoid images overlaid with segmentation mask. A SINAP model was created to segment CRC organoids. D) Graph showing the average CRC organoid area over 5 days between DMSO control and CRCs exposed to compounds (5-FU 5uM, cisplatin 2.5µM, cytarabin 50µM, doxorubicin 15µM, romidepsin 2.5µM, trametinib 5µM). Day 0 refers to images acquired prior to compound exposure. Error bars represent standard deviation between replicate wells.

Viability assay to quantify compound effects

Figure 5. Effects of compounds on CRC organoids. A) Viability assay was carried out on CRC organoids after compound treatment and then imaged. Organoids were stained with Calcein AM for live cells (green), ethidium homodimer for dead cells (red) and Hoechst for all nuclei (blue). Shown here are representative images of organoids. B) The ratio of dead cells (Hoechst and ethidium homodimer positive) for each compound shown. Romidepsin and trametinib treated organoids showed significant increase in dead cells compared to the controls (p<0.001)

Compound-induced phenotypic changes in CRC organoids

Figure 6. Phenotypic changes in CRC organoids. Organoids were fixed and stained with phalloidin after 5 days after compound treatment.

Automation of CRC organoids seeding

To explore the feasibility of automating the workflow, the BAB400 bioprinter was optimized for CRC organoid seeding in a 96W plate. A dome of matrigel mixed with organoids was seeded in the center of each well using the pipette tool (Figure 7).

Figure 7. A) CRCs were treated on Day 2 with trametinib, fluorouracil (5-FU), and staurosporine at 50µM and 25µM, and then stained and imaged on Day 6.
B) Quantification of the average number of live cells in per organoid following drug treatment

Conclusions

• Our results support the use of PDOs for high throughput assays such as compound screening. The availability of standardized, assay ready PDOs provides significant time savings for screens using 3D models
• We show that an AI-based approach can be successfully used to generate robust segmentation for the analysis of label-free biological modes such as 3D organoids.

References

1. Vlachogiannis G, Hedayat S, Vatsiou A, et al. Patient-derived organoids model treatment response of metastatic gastrointestinal cancers. Science. 2018;359(6378):920-926. doi:10.1126/science.aao2774

Download PDF