Assays are integral to successful and cost-effective drug discovery, but optimizing them is no easy task. We asked David Cronk, Director, High Throughput Screening Sciences, Charles River Laboratories, to walk us through some of the challenges in assay development and how his company is solving them; to discuss the role of automation in assay development and the benefits of using a service provider; and to share his insights for the future including thoughts on technological advances that could upend the industry. Finally, as a bonus, he threw in five expert tips on optimizing development of robust and reliable assays.
Biocompare: What is the most significant challenge in assay development and how does Charles River address that challenge?
David Cronk: There are a number of assay development challenges that are faced on a regular basis. The first of these is consistent reagent provision, particularly at scale. This includes both biochemical reagents (be that supply of full length, active protein, or for cellular reagents), such as cell lines expressing the target in a suitable cell background, or sourcing primary cells. These challenges are faced with increasing frequency as the targets diversify, physiological relevance is better acknowledged, and the biology is less poorly understood. In such instances, we will either consider reagent generation in-house or collaborate with a niche provider to address the supply issue.
However, by far the biggest challenge is reproducibility of assay conditions and performance when transferring protocols between organizations. While it is acknowledged across the industry as a whole that recapitulating data between groups is less successful than we would like, it is difficult to pinpoint the exact cause(s), and it may be due to subtle omission within protocols or minor differences in reagents. However, it does present challenges when undertaking validation of assays, and while there is no quick fix where such difficulties are encountered, we would always advocate for collaboration and communication, and we encourage scientific exchange visits as the most time-effective way to overcome such challenges.
Biocompare: What is the role of automation in optimizing assay development?
David: Where multifactorial design of experiments software or matrix assay development experiments are used, automated liquid handlers can be a great asset to simplify the assay setup, and some instruments have built-in wizards to assist with this. But as a general rule, large-scale automation plays less of a role during the early stages of assay development due to the lower volume of data required and the requirement to investigate multiple assay conditions. Having identified suitable assay conditions at a small scale, a key consideration is that not all liquid handlers are the same, and it is not uncommon to encounter very different results when transitioning from a manual approach to an automated system, most likely due to reagent adsorption onto tubing within devices.
Other seemingly mundane but equally important considerations for choice of automation may include the use of labware such as disposable tips or other plasticware from different manufacturers. Different manufacturing practices may affect the quality and the finish applied to the plastic materials, which, in turn, may well change the binding and retention characteristics of the plastic due to a change in the adsorption characteristics, resulting in a difference in observed outcome between small-scale assay development and large-scale automated screening results.
In terms of robustness for high-throughput screening as assays are scaled up, there is an increasing likelihood of encountering issues with variability across a plate, false positives, or false negatives as bulk reagent dispensers are implemented and potentially different incubators are used. It is, therefore, highly important to evaluate assay performance on automated systems and at a relevant scale to identify potential issues as early as possible during the assay development and to investigate different devices if assay performance drops markedly.
Biocompare: What are the benefits of using a service provider to develop custom assays?
David: Most service providers have a sizable team of assay developers who have broad experience with assay platforms and target classes for both biochemical and cellular assays. They are able to draw upon this wealth of knowledge to develop assays and resolve any issues that are encountered along the way. In addition, there are a number of niche service organizations that are highly specialized in particular target classes or assay technologies, and, while their offering is more limited, they play a key role for specialist projects either working alone or through collaboration with a larger service provider. Irrespective of the type of service provider used, the perspective offered by these organizations is likely to be different from your own team, and the focus is to deliver a high quality and robust assay, often within an agreed-upon timeframe.
At Charles River, we see ourselves as an extension of our client’s project team and work collaboratively, yet we are able to view the project objectively and make decisions on whether an assay is fit for the purpose based on the data obtained rather than on emotional investment in the project. This objectivity can be leveraged to good effect, particularly when developing challenging assays where there is a possibility for a no-go decision. Service providers offer flexibility in starting and stopping projects at different phases to allow milestone decision making and, with a pool of assay development scientists available, can assign resources to a project based on the required skills rather than alignment to a particular project team.
Biocompare: Are there any technological advances that you see upending assay development?
David: As an industry, we are great at sharing our successes, but we remain less willing to openly share our failures. As a result, we try assays that should work in theory but do not translate into reality, or we stumble upon inexplicable data caused by interference between targets, detection reagents, or leachates from plastics that should have been previously reported. These patterns and observations are not predictable, render some assays non-functional, and may be unknowingly contributing to reduced performance of others. A mentality shift in the industry toward sharing all results, good and bad, would be transformational in the success not only of assay development but of drug discovery in general.
That being said, as technology progresses, the increasing application of artificial intelligence should be able to interrogate these data and provide predictions on whether the idea, on paper, will translate to reality and point us, with confidence, to a set of near-optimal assay conditions.
At a more technological level, there remains significant emphasis on the biological relevance of assays and how “simple” cell-based assays in particular translate into the tissue microenvironment. With this in mind, the application of 3D cell culture and organoids has yet to find their way into mainstream assay development and hit finding, possibly due to the availability and consistency of reagents but also due to existing equipment not easily lending itself to monitoring biological events within 3D structures. If this could be achieved, preferably utilizing a label-free detection system, we would be able to move toward much more biological relevance.
Five Tips on Optimizing Development of Robust and Reliable Assays
- Understand your target biology, the assay platform you intend to use, and the purpose of the assay. What liability do you have for false positives and negatives from any potential interfering factors and those parameters that may lead you to false conclusions? Are your assay conditions (e.g., protein concentration, incubations time, etc.) in accordance with the assay aim?
- Keep your initial experiments simple. Demonstrate that your assay concept works before embarking on more complex experimental designs. Be open-minded when analyzing the data, and examine the raw data values; do not rely on normalized or ratiometric values alone to guide decisions.
- Utilize Design of Experiments software if it is available to you. If it isn’t, evaluate parameters in combination using a matrix of conditions, including suitable replicate numbers, to identify where there may be interacting factors. When considering your experimental design, keep in mind the end goal of the assay; the requirements for a large compound number high-throughput screen are quite different from the requirements of an assay in which a limited number of compounds will be tested across a concentration range.
- Consider the long-term stability of reagents, especially when dealing with cells. Even minor variations in your cell-handling conditions can lead to marked differences in the final assay data, and ensuring consistency in cell density and passage frequency can make all the difference to assay performance. The use of frozen cells can often help, assuming the assay system is amenable to this.
- Track your data as you progress through the assay development process to spot trends in signal associated with operators and assay days. There can be a tendency to optimize based on size of assay signal and standard parameters of assay performance. While it is always encouraging to obtain strong statistics, these should not be obtained at the expense of biologically relevant assay conditions.