Part Two: Approaches to Finding Good Hit Compounds - The Old and the New

Written by SelectScience Guest Editor, Dr Peter Simpson, AstraZeneca.

In this second part of this article, Peter Simpson discusses other approaches to high throughput screening for lead identification, and possible ways of supplementing current techniques and methodologies using phenotypic screening as the main example. In the first part, Peter summarized the limits of high throughput screening for ‘hit’ identification and the possible solutions to these failures.

From various meta-analyses that have been published over recent years, conventional wisdom seems to have come round to the view that high throughput screening has ‘failed’ in its promise of delivering better, faster medicines by rapid identification of good chemical starting points. New (or indeed old) replacements for high throughput screening have been proposed by various opinion leaders. A substantial movement has emerged with the view that ‘phenotypic’ screening is the solution, to at least some of the problems that have made it challenging to translate high throughput screening activity into successful lead molecules.

Phenotypic screening was discussed in part one – so, is there any reason why a phenotypic approach to finding ‘hit’ compounds should not translate well to mammalian cells and targets?
Well, it is easy to underestimate the amount of time, and effort, that will be required to move from activity in a phenotypic assay to a convincingly identified molecular interaction that drives the disease biology, and then to identify and deliver selectivity over other related molecular targets. Some advocate that precise molecular selectivity is part of the problem, and poly-pharmacology (action of drugs against multiple targets) approaches naturally derived from phenotypic assays will prove to be more impactful in the genuine cellular context. This may sometimes be post hoc justification for compounds that are unavoidably non-selective. In other cases, it is true that just inhibiting single molecular targets can enable a cell to easily bypass this molecular target, read just expression levels of other enzymes that do a similar task, and proceed as before. (This underlies why combination screening, in this example using microfluidic liquid handling systems, is another emerging trend in oncology and other diseases; but that is for another article.)

DNA encoded library technology
Other approaches are also looking to replace or supplementhigh throughput screening. For example, DNA encoded library technology (in which small molecules are tagged with DNAs that serve as tracking devices, The Scientist) has seen huge levels of investment from multiple pharmaceutical companies either in internalizing, or accessing, this approach from a variety of venders and companies. The potential here is to greatly increase, from millions to billions, the number of compounds screened to interact with the target protein in a conventional, reductionist, affinity assay. As well as a high diversity of compounds screened, a major advantage of this approach is that it is possible to run multiple assay conditions in parallel, to only select compounds with the appropriate pharmacology, selectivity and other features as hits. This is ‘ultra’ high throughput screening, and relies on a binding event rather than efficacy event, and so is arguably more reductionist than most efficacy-based high throughput screening.

'Right approach to the right target'
So, we could see high throughput screening being supplanted either by a more reductionist, higher throughput approach such as encoded library screening, or by a more complex but more biologically relevant approach that is phenotypic screening.

The true solution should be to apply the right approach to the right target. It is, however, unfortunately still not always easy to see which approach matches the right target based on the current level of knowledge of success rates for phenotypic, and DNA encoded library assays in particular. We do have extensive data sets available on which targets and types of assays are successful within high throughput screening, though this information is not always made publicly available. Pharma companies should have sufficient knowledge to apply that approach to the right targets. We can legitimately hope, maybe expect, that the emerging approaches including phenotypic screening, DNA encoded libraries, and other new technologies in the small to large molecule space, can deliver leads for targets for which high throughput screening is not the right solution. We have more tools in our hit discovery toolbox - we must learn how to use them wisely.