Download PDF

2D cell lines are poorly predictive and ineffective at detecting therapeutic responses

In drug discovery, the focus has shifted recently towards the development of compounds targeting the so-called ‘cancer stem cell’ populations within tumors. Non-specific, “Standard of Care” chemotherapeutics target rapidly dividing cancer cells in order to reduce the tumor bulk; the effects of which are usually to drive tumor regression in the short-term, albeit with greater toxicity (side-effects) and a high chance of patient relapse. It is now widely understood that, in order to permanently ablate tumor growth, the initiating cancer stem cell population must be removed or inhibited. In the patient, this could manifest as a relatively small impact on overall tumor size initially, but a longer-term more effective treatment caused not by cell killing, but by a more subtle change in the behavior (phenotype) of the cells within the tumor (Fig 1A). The search for such targeted compounds has led to demand for a screening platform that better predicts compound efficacy; while historical drug discovery has relied heavily on the predictive power of 2D cancer cell lines, their lack of cellular heterogeneity and relevant phenotypic behavior leaves them largely unsuited for the study of CSC (cancer stem cell) inhibitors (Fig1B.i).

Patient-derived tumor organoids – structures retaining intra-tumoral complexity and crucially, stem cell function – are now gaining increasing merit as predictive in vitro models in the drug discovery field, with the potential to reduce compound attrition rates and development costs, ultimately increasing the number of successful compounds available for use in the clinic.

Proof-of-concept study for the use of 3D organoids to detect response to a stem cell-targeted therapy

A new study exploring the application of patient-derived organoids in the study of novel inhibitors of stem cell activity has recently been published in the journal PLoS One (Badder et al., 2020). In brief, the study describes the derivation of a novel set of patient-derived colorectal cancer organoids* and the investigation of their response to Wnt pathway modulation by a number of tankyrase inhibitors (TNKSi). The work relies on a range of analysis techniques, but highlights one quantitative method in particular as having the potential to greatly enhance the high throughput prediction of compound efficacy in pre-clinical testing.

Figure 1. Comparison of true patient treatment outcomes with those predicted by in vitro models. (A) Tumors treated with standard chemotherapy initially regress, leaving a small population of cancer stem cells to re-establish the tumor bulk in the long term. Tumors treated with a stem cell specific therapy may initially appear static, but over time regress as their source cells die out. (B) In vitro model systems such as 2D cell lines and 3D organoid cultures give different predictions of stem cell therapy outcomes. 2D cell lines lack the heterogeneity to identify stem cell specific effects, and fail to predict in vivo responses accurately. 3D organoids have the ability to show response to treatment, but outcomes are determined more accurately by multiparametric phenotypic measures than metabolic readouts alone. Key: orange cells denote cancer stem cells, blue cells denote bulk tumor (non-stem) cells, green cells denote phenotypically altered cells following treatment.

Using ‘single-parameter’ biochemical assays, the study identifies the differential responses of a range of organoid lines from varied genetic backgrounds to a novel TNKSi. Partial efficacies are detected in specific organoid lines using ATP measurements, in a way that notably, would not have been possible using more homogenous 2D cancer cell lines (Fig 1B.i). Such ATP-based readouts of the metabolic activity of cells – as a proxy for cell viability – traditionally used to assess 2D cancer cell line responses – have remained a popular choice in the early adoption of 3D organoid models into drug development pathways, owing to their low cost and compatibility with high throughput systems.

However, this study highlights limitations to the application of such measurements specifically when considering targeted treatments. While informative enough to delineate broadly the organoid lines that may be sensitive to treatment, their ‘aggregate’ nature – that is, the readout of a single value from one well of a large number of organoids, each comprised of a varied proportions of cell types – leaves them lacking the sensitivity required to detect subtle, but crucial, changes within specific cell populations and thus greatly reduces assay windows in studies of targeted compounds. For example, a stem cell targeting compound that drives terminal differentiation may leave the overall number of ‘viable’ cells unchanged, or only marginally reduced (Fig 1B.ii). This would be interpreted as a partial or limited effect, and the drug may well be excluded in early drug development studies. In reality the cells are altered phenotypically to the extent that the organoid is entirely prevented from progressing any further, and in vivo, it’s corresponding tumor would ultimately regress (Fig 1B.ii)

Similarly, biomarker analysis has been widely used in 2D drug screens where the target pathway is well characterised; here, analysis of total B-catenin levels and gene expression of its downstream targets was used as a proxy for Wnt pathway activity following TNKSi treatment. Although this study observed a correlation between some specific stem cell markers and B-catenin levels in responsive organoid lines, the two readouts did not always correlate with a functional response. In a largescale screening setting, failure to choose those markers that reliably predict functional response every time could risk introducing false negatives or false positives into analysis quite quickly.

Therefore, the findings here argue that a more complex model requires a more comprehensive method of analysis, capable of capturing and integrating the entire range of changes that may occur in response to treatment.

By implementing 3D image-based analysis workflows, the study quantified \~600 different morphological features from individual organoids following treatment with TNKSi (including those related to shape, size and branching), and identified those that best described compound response.

This morphometric approach determined responses that, importantly, mirrored the trend of those seen in ATP assays, but by virtue of the aggregation of multiple parameters, was less affected by noise and variation, enabling alterations in growth and morphology in response to TNKSi to be detected much more sensitively than by conventional single parameter readouts.

This study indicates that while biochemical studies may still have a place in the early stages of the drug discovery process in detecting compounds warranting further investigation, combining biochemical assays and morphological analysis could be the key to unlocking the full potential of organoids in predictive drug testing on a much larger scale.

Overcoming barriers to the use of organoids in high throughput screening

Prior limitations to the widespread integration of organoids into high-throughput assays have included limited scalability and reproducibility due to manually intensive and costly culture processes. However, key developments in bioprocessing technology by Molecular Devices now allow the commercially scalable generation of consistent, reproducible batches of organoids, offering the model as a truly feasible option to the drug discovery market.

Going forwards, the adoption of organoids into highthroughput workflows will demand the availability of highly sensitive methods of analysis that can take full advantage of the complexity of information that organoids offer, in a cost efficient manner. The findings here are a clear demonstration of the improved power of multiparametric phenotypic analysis as a measure of stem-cell related drug response and are therefore likely to be of great interest to the drug discovery sector, promoting the adoption of organoids into HTS in the longer term.

How can Molecular Devices help with HTS in 3D?

Molecular Devices’ patented bioprocess enables the large-scale expansion of reproducible, validated batches of organoids compatible with high-throughput screening applications:

• 3DReady for plating straight into desired format
• Full protocols
• Technical support
• Large-scale expansion service with custom vialling

Download PDF