“My thought was, well, why are we screening at just one concentration?” he says. “You get a lot of false positives, or you miss things.” Instead, Inglese and his colleagues developed a new method, called quantitative high throughput screening (qHTS), which can screen across up to five orders of magnitude difference in the chemical concentrations. The result is a dataset of titration curves which allows researchers to begin to see pharmacological trends such as the IC₅₀ and EC₅₀ within their initial screens.

In one recent application of qHTS, Inglese and his colleagues married qHTS a biological assay of the malaria parasite, Plasmodium falciparum, developed by Inglese's colleagues at the National Institute of Allergy and Infectious Disease (NIAID) that can be used with their low volume 1536-well microtiter plates. By using this strategy to examine the effect of library compounds on parasite viability, and because of the genetic variability of parasites from different locations, the data has helped the researchers pinpoint the basis of some of these response differences and look at resistance factors to particular drugs (1,2).

Chemistry to follow up on the screen

Chemistry is also important after the screen. An initial high throughput screen produces a number of hits, molecules that act on a target in some way. But this is just the beginning; these molecules need to be validated and optimized, and that's where medicinal chemistry centers play a role in the Molecular Libraries Probe Production Centers Network (MLPCN): Aubé directs the Kansas Specialized Chemistry Center and Vanderbilt University has its Specialized Chemistry Center for Accelerated Probe Development

Once that initial screen is complete, Aubé and his colleagues join forces with a team composed of researchers from the screening centers and assay developers to decide which initial hits should be pursued using confirmatory assays. It is then that the medicinal chemists will follow up on those hits and construct structure-activity relationships. “There are milestones and targets of potency and selectivity and physical properties that are associated with each of the probes,” Aubé says.

Probe molecules can also feed back into investigations of the molecular mechanisms that generated the hits in the first place. Many of the screens are phenotypic: researchers are looking for a desired cellular response and identify hit molecules that generate that response. But most researchers would also like to go beyond phenotype to better understand the biochemical pathways that generated a particular response.

After all this work is complete, the results are published as probe reports on the MLP website and purified samples are kept available for other researchers to use. One probe report published by the group last year details a molecule with a novel structure that blocks the activity of CDC42, a GTPase that regulates the cell cycle. Since late February, Aubé estimates he's received a dozen requests for this particular probe.

“As the network has gone forward, we've been making a deliberate effort to ensure that we're providing not only compounds that are going to inform basic biology, but, when it's reasonable to do so, also provide compounds that might be the first step of a lead compound for drug discovery or translational research.”

Library bias and enhancing diversity

In the early days of high-throughput screening (HTS), a lot of potential hits turned out to be artifacts. Brian Shoichet and his colleagues at UCSF saw similar patterns in their own structure-based screening work and started to wonder if the problem was with the methods being employed or in the molecules that were being used for screens.

Around the same time, Shoichet became aware of articles estimating the vastness of chemical space, which led him to another question: with the limited number of compounds available, why does high throughput screening ever work at all? In 2009, Shoichet and his colleagues published a computational analysis revealing that existing libraries are biased toward biogenic molecules (4). Molecules in existing libraries tend to be particularly good for targeting G-protein coupled receptors, ion channels, and kinases. Therefore, if researchers blindly increase chemical diversity, the likelihood of hits might actually get worse. A better strategy is to think in terms of the molecules and scaffolds organisms likely have in their environments, such as natural products.