Jump to content
Powered by

Drug screening - higher throughput, quicker and more effective thanks to automation

Over the last few years, automation has revolutionised the search for pharmaceutical compounds. Using methods such as high-throughput screening or high-content screening, it is possible to analyse thousands of molecule activities very quickly. Screening for effective compounds is now done virtually with computers. This trend of increasing sample throughput is set to continue.

Screening for a substance (blue/red) that binds to the body’s own proteins (grey). (Photo: Boehringer Ingelheim)
The expression “looking for a key that fits the keyhole” crudely describes the search for a specific, pharmaceutically relevant protein for the treatment of a particular disease. ’Screening’ is usually the first step in the drug discovery process, involving the testing of millions of substances for their ability to bind to a particular disease-causing protein (target) thus blocking the effect of the protein as well as inducing a curative or alleviating reaction in the body. The screening process specifically focuses on collecting information on the desired properties of molecules that bind to a particular target and whether they need to contain certain atoms or not. For example, a screening process can investigate the optical absorption, intensity of fluorescence and the polarisation in a particular test volume. Up until the 1990s, the search for active drug ingredients was done ‘manually’ and usually took several months or even years to finish. Since the 1990s, pharmaceutical companies have set up huge robot-run facilities enabling them to fully automate the tests for the interaction between the target molecule and the synthesised substances. High-throughput screening (HTS) is the core of this process and enables the very rapid analysis of hundreds of thousands to millions of substances.

High-throughput screening – rapid ‘hit’ identification

High-throughput screening usually involves microtitre plates with 384 or 1536 wells that are analysed 100 percent automatically. The process includes sorting, portioning, mixing and measurement. During the procedure, liquid handling stations add exact amounts of different reagents to the wells at a precisely determined point in time. The molecular interaction between the individual components is subsequently determined using an optical reader. The entire process, from the handling of the microtitre plates to the acquisition of data involves computer software that is becoming more and more effective. State-of-the-art HTS systems have a throughput of at least 10,000 samples per day; ultra HTS systems have a capacity of between 100,000 to 200,000 substances per day. In the majority of cases, one in two hundred or one in a thousand substances is tested positive (hits).
Modern robot facilities enable high-speed substance screening. (Photo: Nycomed)
Modern robot facilities enable high-speed substance screening. (Photo: Nycomed)
Pharmaceutical and biotech companies have established comprehensive collections of all kinds of substances for HTS. These substances originate either from traditional substance libraries, combinatorial chemistry libraries, or are natural substances or derivates thereof. The most frequently used assay types are receptor-binding studies or enzyme activity determinations as well as second messenger and reporter gene investigations. Potential targets include kinases and other enzymes (e.g., phosphatases, proteases, nucleases), cell surface receptors or ion channels.

High-throughput screening has become a key technology in the drug discovery process. More than 60 per cent of Boehringer Ingelheim’s fully developed lead structure optimisation programmes are based on the success of HTS. Over 2.5 billion US dollars are spent annually on HTS products and HTS services in the USA alone. Large pharmaceutical companies invest up to 30 million euros annually in screening technologies; the major proportion of these investments goes into the development of new assays. According to Dr. Werner Stürmer, head of Early Discovery at Nycomed, clients usually pay 0.50 to 2 euro per assay point.

Additional information gained with high-content technology

High-throughput screening traditionally involves biochemical assays to analyse substance-target binding or enzyme inhibition, for example. However, besides the many false-positive hits, the method does not provide any information about the interaction of the substance with other elements of biological systems.

In contrast, high-content screening (HCS), which involves cell-based assays, provides further information on this aspect. While HTS usually only tests one parameter, HCS simultaneously and quantitatively assesses several parameters on the cellular and subcellular level, including absorption, permeability, selectivity or metabolic properties. Automated HCS is based on fluorescence microscopy involving labelled cells. Cells are sown on microtitre plates, they are then specifically treated and fixed with a substance. Detection involves fluorescent tags, for example fluorescent antibodies or green fluorescent protein. Images are then automatically taken of the stained cells and subsequently quantitated using specific software algorithms.

HCS systems are based on automated digital imaging technology, in combination of high-quality bioinformatics software for the analysis and storage of the huge amount of complex data. HCS technology is suitable for testing up to 60,000 substances per day.

Virtual search for geometric similarities

Computer technology and bioinformatics are becoming more important in the determination of lead structures. However, although the prediction accuracy of computer-based screening is still far from being comparable to the accuracy of physical grid investigation, computer-based methods help to save time and money and contribute to efficiently increasing the number of screening substances in a given project. In general, the cost of screenings, including the validation of hits, ranges between 35,000 and 700,000 euros, depending on size and complexity. According to Dr. Herbert Köppen, head of Computational Chemistry at Boehringer Ingelheim in Biberach, the cost of the classical physical testing of substances is up to ten times higher than for a ‘virtual’ screening.

One of the major tools of virtual screening, in which biologically effective substances are identified from a large number of substances through arithmetical selection of compounds or libraries, is molecular modelling. Here, the target and the substances to be tested are simulated on the computer in order to identify the pharmaceutically active structure. As the molecules of a pharmaceutical substance and those of the body’s own protein system that are targeted by the substance need to be spatially congruent, molecules are tested to find out which ones are able to bind to the target.
Molecular modelling is an important tool in the drug screening process. (Photo: Nycomed)
The methods used are graphics algorithms (for the visualisation of molecules), geometrical calculations (molecular surface) or methods such as docking, in which ligand structures are docked and slotted into a binding pocket of a protein using 3D structure information and the calculation of binding energies according to the key-lock principle. Large databases and substance libraries are the main foundation of computer-assisted drug discovery. Large pharmaceutical companies have specific search methods that enable the easy screening of a billion virtual structures on the computer. According to Dr. Andrea Zaliani from Nycomed’s Molecular Modelling department, the world largest physical substrate collection contains about 50 million structures.

Automation requires multidisciplinary cooperation

In many companies, molecular modelling is a helpful tool for HTS departments. It is common for the search for effective substances to be done using HTS and virtual screening, in which the latter actively helps identify compound classes and propose target-oriented commercial compounds, which will then be acquired and tested. Therefore the final result is that the identified lead structures do not necessarily originate from a single screening approach, they may also originate from virtual or structure-driven approaches from medical chemistry.

As a result of the increasing use of robots in drug research, more and more interdisciplinary cooperation is needed in screening processes and technologies. The biology laboratories of large pharmaceutical companies are increasingly working with engineers and physicists to work out automated test protocols. These efforts are supported by IT specialists who are developing computer-based approaches or assisting in the acquisition and processing of data.

Inclusion of innovative screening tools

In order to accelerate throughput in HTS processes, pharmaceutical companies are working hard to further miniaturise the assays, for example to achieve a higher density of microtitre plate wells or to use biochips on which proteins, peptides or nucleic acids are immobilised. The gradual expansion of substance databases puts pressure on companies to achieve higher throughput rates. At the end of the 1990s, substance libraries in HTS departments contained on average 50,000 to 350,000 compounds whilst modern libraries contain between 1 and 1.5 million compounds. “In general, the libraries are expanded by up to 10 per cent per year, with a strict focus on useful structural diversity,” said Dr. Werner Stürmer.

In addition, pharmaceutical companies are currently working hard on the use of new, made-to-measure screening strategies in order to obtain high quality results that lead to a reduction in the number of false-positive hits, for example. Major trends in automated drug discovery are technologies such as auto-patch clamping, a method for the measurement of ion channel activities. Ion channels are part of the cell membrane and they are the third largest group of targets for pharmaceutical drugs after enzymes and receptors. Almost one tenth of all projects in the pharmaceutical industry are based on ion channels because patch-clamp robots enable the rapid screening of voltage-dependant and ligand-activated ion channels as well as the testing of different substances on a single cell to be carried out. Such patch-clamp automates for high-throughput measurements are currently in very high demand in drug research.

It is believed that future high-throughput methods will focus more on the integration of new methods focusing on additional target classes such as protein-RNA/protein-DNA interactions or transmembrane receptors.

mst - 25 Sept. 2008 © BIOPRO Baden-Württemberg GmbH
Website address: https://www.gesundheitsindustrie-bw.de/en/article/dossier/drug-screening-higher-throughput-quicker-and-more-effective-thanks-to-automation