Application Note
Understanding organoid morphology: a study to assess organoid size and cell count from 3D image analysis
Summary – What are Patient-Derived Organoids (PDOs?)
- Derived from normal or diseased patient biopsy tissue
- Grown in 3D in a protein gel with liquid feed
- Formed from multiple cell-types (as are organs in the body)
- Miniature organ replicates (e.g., “mini guts”)
- Fully representative of human biology
- Patient “avatars” for use in drug discovery
- Grown in a bio-processor to create millions of standardized copies per batch
Introduction
PDOs and drug discovery
Compound library screens using 2D monolayer cancercell lines are commonly performed in the pharmaceutical industry, to identify molecules with therapeutic potential. Multiple analogues and modified derivatives are generated and tested before candidates progress to clinical trials. This is a lengthy and costly process with a poor success rate (less than 5% for oncology). This is partly due to the failure of the current generation of preclinical cell-based assays to accurately predict clinical efficacy.
3D patient-derived organoids (PDOs) have improved predictive power and have the potential to replace 2D assays in drug discovery (Vlachogiannis et al., 2018 Science 359, 920–926, van de Wetering et al., 2015, Cell 161, 933–945). However, their utility is currently limited by the lack of standardised assays with proven applicability for the pharmaceutical industry and the difficulty in producing PDOs in sufficient quantity, with the required quality and reproducibility. The latter has been addressed by Molecular Devices, through the development of a unique and patented industrial bioprocess for the largescale expansion of PDOs. This proprietary bioprocess generates highly standardised PDOs at scale, enabling applications such as high-throughput screening for drug discovery.
Organoid morphology in relation to pathology
The challenge for organoid users is to fully extract, exploit and understand the increased complexity of 3D in vitro models using imaging for example. Many of the practical limitations have already been overcome by recent advances in microscopy, leading to improvements in 3D image resolution (Figure 1), speed of imaging, data acquisition and dedicated software for high-content, high throughput assays. This technology can be used to quantify changes in gross organoid morphology, such as size and shape and also detect subtle, cellular, drug-induced alterations. (Figure 2, B and C are visually different compared to the untreated control A.) Multi-parametric, image-based analysis can lead to an understanding of the relationship between the morphology of 3D PDOs and the underlying cell biology (Badder et al. 2020, PLOS ONE).
Analysis and standardization are central to the adoption of PDOs as a platform for drug discovery. In order to demonstrate that 3D imaging can produce relevant data, Molecular Devices carried out a simple demonstration of the acquisition of visual and numerical data with subsequent analysis, to quantify the supposition that the number of cells per organoid increases with size. Many relevant, alternative properties of PDOs (such as nuclear shape, size and number of lumens) could be used to compare untreated and compound-treated PDOs. The resulting analyses could be used to predict clinical efficacy and ‘mode of action’ and to establish a quality framework for PDO use.
Figure 1. Colorectal tumor organoids, stained with Hoechst to detect DNA in the nuclei (blue) and phalloidin to detect F-actin in the cytoskeletal structure (yellow), showing the lumen(s) of the organoids. Images were acquired at the National Physical Laboratory using M Squared Lasers’ Aurora Airy Beam sheet imaging system.
Comparison of the morphology of untreated and treated patient-derived colorectal cancer organoids
Figure 2. Images of ISO 50 organoids, ± 5 days exposure to colorectal cancer targeting drug treatments. This line was derived from a rectal tumor, Duke’s stage C2, genotypically characterised by commonly reported key driver mutations in APC, TP53, KRAS, SMAD2 and SMAD4. Following treatment with 5 Fluorouracil (5FU) or Trametinib, there is a visual difference in organoid shape and size. A: Untreated control B: 625nM 5FU C: 6.25nM Trametinib. The scale bars represent 100 μm
Experimental design and data acquisition
Methods
Colorectal cancer organoids were cultured manually, seeded into multi-well plates and allowed to grow for up to 8 days. Every 24 hours from 3 to 8 days, a well was fixed and stained with Hoechst and Phalloidin (as illustrated in Figure 1). The organoids were imaged with a confocal high-content screening solution at 40X magnification, with 50 image stacks taken at 3 μm step-intervals per well.
Data analysis
Imaging software was used to analyse the z-stack images with 3D reconstruction. An ellipsoid approximating the shape of each organoid was fitted and the number of stained nuclei within the ellipsoid were counted. Cell numbers (equivalent to the number of stained nuclei) were plotted against the ellipsoid medium axis length, enabling the average number of cells per organoid to be calculated. (Figure 3).
Figure 3. A. A nonlinear regression of the data shows the increase in the number of nuclei per organoid with increasing size, in line with the length of time in culture. B. An image of nuclear segmentation for colorectal cancer organoids grown for 5 days. The different colours show individual cells. The scale bar represents 50 μm.
Results
As expected, the data analysis showed that in culture, the number of cells per organoid increased over time, as did the organoid diameter.
The average ellipsoid medium axis length can also be used to calculate the area of the maximum 2D projection for each organoid, which shows a linear relationship with the number of cells per organoid as shown (Figure 4).
These particular readouts could be used for quantifying the cell-killing effect of drugs or to assess specific ratios of different cell types within an organoid. It is also a factor to be considered in manufacture and experimental design.
Figure 4. Individual organoids grow at different rates as shown by a higher variability in both the size of the organoids after 8 days in culture and the number of cells per organoid. Data must be acquired and analysed for each different organoid line to make correlations between organoid size, number of cells and specific morphology.
Conclusions & summary
Patient-derived Organoids are the drug discovery platform of the future because of their greater physiological and clinical relevance as compared to existing 2D models used in screening. Importantly, this work demonstrates the principle that morphological parameters can be assessed and quantified in a high-throughput format.
Once key morphometric signatures for clinical efficacy have been defined, 3D imaging has the potential to revolutionize drug discovery, leading to the early selection of hit compounds that will progress through to the clinic. Assays for changes in organoid morphology following drug-treatment have the potential to identify treatments targeted to specific cell-types such as cancer stem cells, that would be missed in the metabolic assays used currently, since there is no bulk cell-killing effect.
Greater success rates in drug discovery will reduce the cost of drug development and increase the number of effective drugs on the market. Directed treatments result in less off-target toxicity, reducing the length of hospital stays, improving survival rates and the quality of life for cancer patients.