Cell Painting: The Colorful Future of Drug Discovery
This panel Q&A was written in collaboration with PhenoVista and highlights their unique service offering, Cell Painting. Learn more about Cell Painting from Jeff Irelan, PhD, Sr. Director of Commercial Operations, Matt Stoltz, Sr. Director of Business Development and James G. Evans, PhD, CEO, as they share their insights on how this service can aid your drug discovery process.
What is cell painting?
- Cell painting is an unbiased, phenotypic drug-discovery approach that uses predictive model systems to evaluate therapeutic candidates. The “painting” part refers to the application of fluorescent probes specific to various cell components or compartments. For each compartment, high-content imaging allows for quantitative assessment of multiple features, such as signal area, intensity, puncta count, and shape.
We use this technique to assess the behavior of cells in response to treatment with various test articles. The test articles may be small molecules, biologics, or even genetic manipulations, such as siRNA knockdown or CRISPR editing through a number of quantitative readouts.
A key part of Cell Painting is the use of sophisticated bioinformatics approaches, including machine learning, to classify the test articles according to their corresponding cellular responses.
Where is cell painting useful in the drug discovery process?
-Cell painting is usually employed in preclinical and early discovery programs. The unbiased profiling of cellular responses can provide novel mechanistic insights. The approach can be especially useful for drug repurposing (exploring new indications).
- We compare the cellular response profile of test articles to reference treatments of known mechanisms, so hypotheses can be generated as to the test molecule’s MOA. This can enable better understanding of both on-target biology and potential off-target toxicities.
What are the dyes you use, and what data can they provide?
- The original or canonical cell painting palette consists of 5 “paints”:
1. Hoechst stains nuclei for nuclear morphology features, as well as cell count/overt toxicity via cell loss
2. SYTO14 stains nucleoli and cytoplasmic RNA
3. Concanavalin A stains the endoplasmic reticulum
4. Phalloidin and WGA are combined to stain cytoskeleton, Golgi and plasma membrane
5. MitoTracker FM stains mitochondria
- We often customize the palette to meet specific client needs. For example, we can substitute markers specific to a particular target, phenotypic readout, or cell type for one or more of the canonical readouts. In many cases, these are antibodies, which usually require some degree of optimization to be utilized in the cell painting context.
What library/references do you compare test compounds against?
Every cell painting study we run includes technical controls for each marker plus reference molecules specified or pertinent to client biology. We are building on a proprietary database by constantly adding new reference molecules.
Some clients help us expand the database by providing shared access to reference compounds or libraries they provide or nominate, in return for a discounted screening price.
What analysis is available, and how are the data presented?
- This depends on the scope of the project and the questions being asked. For large datasets, principal component analysis (PCA) is a method that can reduce a large set of input features to a smaller one, while accounting for most of the variance in the data. This approach can provide an overview of the results. For a small set of test articles where MOA identification is the key question, the most useful output may be a rank ordering of “nearest neighbors” comparisons based on Euclidean distance (the shortest distance between the mean of a given test article relative to a selected reference compound). This would typically be presented for the most informative features in the dataset. Finally, for select features of interest, the dose-dependent results across test articles and select reference molecules may be informative. We work collaboratively with clients to deliver the most useful data to answer the most relevant biological questions.
- Also, keeping in mind that “a picture is worth a thousand words,” we provide representative images for all the key test conditions and readouts. The images can be quite striking with the multi-color palette and often are used by clients for their own marketing and fundraising purposes.
What cell types do you use for these screens, and why?
- The canonical cell painting methodology was developed using simple-to-handle and easily imaged tumor cell lines, such as A549 lung carcinoma cells. We have performed many screening runs using A549s and the canonical palette described above, and many clients find this mode useful, especially for utilizing existing datasets for comparative MOA determination. For some purposes, more biologically relevant and predictive model systems such as primary or iPSC-derived cells are more appropriate, and we have performed several campaigns using those cell types, driven by specific client needs.
- As an aside, we have a non-exclusive partnership with FUJIFILM Cellular Dynamics, Inc. (FCDI) to utilize their iPSC products for client studies and are a preferred provider of services for North America.
How are PhenoVista’s capabilities different than other offerings, or from someone going out, buying a kit, and doing it themselves?
- Although the canonical cell painting methodology is very well established and not terribly tricky within the universe of high-content imaging applications, getting it up and running can be challenging. Even with off-the-shelf staining kits, there is usually a significant amount of assay development or transfer work.
- On the analysis side, many or most of the very large number of features that are quantified in canonical cell painting are usually not informative to the questions our clients are asking, so we provide more efficient analytical approaches to facilitate data interpretation. Each of the aforementioned customized painting palettes and cell models would also add significant layers of complexity, risk, and optimization time to develop the assay. By outsourcing to us, our clients benefit from our prior experience in setting up and executing a wide variety of cell-painting assays. This reflects the over-arching goal of PhenoVista: to ‘democratize’ high-content imaging by turning the complexity of high-content imaging into interpretable data for day-to-day, drug-discovery work.
Connect with Phenovista.