Finding the Right Assay Conditions
Developing phenotypic assays that model human disease is a heavy lift. Our understanding of disease biology is limited. And there are so many variables, it’s hard to know where to start.
I’ve been very fortunate to have been involved in developing hundreds of human primary cell-based models of disease and tissue biology. Working in a small start-up focused on drug discovery we had the luxury of doing exploratory research to test many different assay conditions.
We began with a simple hypothesis: that assays using human primary cells would be more predictive of human outcomes. We chose primary cells for their high level of physiological relevance (although they are not without their challenges). After many iterative cycles of testing and learning, we did succeed in developing some useful assays (many of which can be found here). However, the path was never direct and there were many surprises along the way.
Unexpected benefits of co-cultures
One of the first assays we developed was a co-culture of endothelial cell and peripheral blood monocytes (PBMC). We weren’t trying to develop a physiologically relevant disease model. Instead, we were interested in an assay to screen for inhibitors of T cell activation. Conventional assays that measure T cell activation (using antibodies to CD3 and CD28) were expensive and variable. Our concept was to use an alternative stimulus (bacterial-derived superantigens) and endothelial cells in the co-culture to detect activated T cells. This assay turned out to be more economical and displayed superior performance characteristics. And we were able to detect and distinguish not only mechanisms involved in T cell activation but also endothelial cell mechanisms (Kunkel et al., 2004).
The power of combinations
We initially explored combinations of stimuli (e.g., growth factors, cytokines, etc.) as a way to generate assays for screening multiple pathways at once. We also recognized that activation of multiple pathways together can activation states that are more representative of disease tissues. Treating vascular endothelial cells with a combination of cytokines produces an expression pattern of leukocyte endothelial cell adhesion receptors that more accurately matches the pattern observed in chronic inflammatory diseases. Remarkably, we discovered that particular combinations of stimuli produced assays with particularly good performance characteristics (unusually high reproducibility). Finding these combinations was often a lengthy process.
The value of designing for scale
In many ways, our commitment building assays that could be run at scale contributed to our success. Only with high throughput could we have tested the number of options and variables needed to “find” the right assay conditions. During the assay development process, high throughput facilitates testing of large numbers of variables and conditions. It is also important for assay validation, which when done appropriately involves testing of multiple phenotypic assay standards at multiple concentrations and replicates.
For those of you developing phenotypic models of human disease, I wish you the best of luck. Don’t be afraid of wild ideas. You can’t win if you don’t play!
Photo by Dmitry Antropov on Unsplash.
