• This article provides an overview of the current status of the high-throughput screening (HTS) landscape, with a special focus on the presentations at the IBC ScreenTech World Summit 2003 (that contained a number of conferences on related topics, held in parallel), which was held recently in San Diego, CA.
• In the current article, we present and discuss some of the HTS metrics (quantitative parameters characterizing the current space) that the industry currently accepts as well as some of the drivers therein, and a landscape of emerging themes in terms of technologies and their applications in various sectors of the HTS space.

Introduction
Many have characterized HTS as no longer a bottleneck in the drug discovery and development process. Indeed, the new technologies and products in the sector have driven productivity in HTS and have reduced the bottleneck in screening, but the space still has a number of challenges that new and emerging technologies seek to address.

A key component of the need for new technologies and approaches in HTS is driven by the emergence of a large number of targets from genomics. Previously undiscovered targets are now in the realm of the screening enterprise, and pharmaceutical and other drug discovery companies are looking to extract value from this treasure trove. Unfortunately, however, many of these targets are refractory to screening using traditional approaches, and therefore, new methodologies need to be devised to address them in the context of drug discovery.

Good examples are the increasing numbers of G protein coupled receptors (GPCRs) as well as ion channel genes that have been uncovered in the genome. These target classes are notoriously difficult to screen using traditional methodologies. In the case of GPCRs, the screening needs to take place in a cell-based format in order to extract the most meaningful biological result (since GPCRs are studded in the cell membrane and function physiologically in concert with a number of accessory proteins). Therefore, for each GPCR, ideally, a cellbased assay is required, and it is difficult and costly to build cell-based assays that are a bona fide reflection of the in vivo situation. In the case of ion channels, another target class of value in drug discovery, the current methodologies for screening are of exceedingly low-throughput and hence difficult to scale up for HTS. Furthermore, with ion channels, it is key to characterize them fully with a number of different parameters, and therefore, their screening is not a simple binary read (signaling or blockage, for instance). This complicates their analysis, yet their economic importance (their involvement in a large number of physiological and pathological processes) is unquestioned. Therefore, the industry is striving to develop technologies to address ion channel targets at high-throughput and in a high content manner.

Quantitative HTS Metrics from the Pharmaceutical Community
In order to frame the current status of the HTS industry, it is important to characterize some of the metrics that govern screening campaigns within pharmaceutical companies. We present in this section some metrics from the recent HTS conferences.

• From conversations with the industry participants, we estimate that an average weekly throughput of 80,000 wells or more is classified as HTS in the average HTS community.
• In addition, the average number of molecular targets (such as GPCRs, kinases, ion channels, nuclear hormone receptors, etc.) that are screened is about 28-30 per year.
• In terms of the size of the screening libraries (of small molecule compounds), typical medicinal chemistry libraries contain hundreds of thousands of compounds. Also, they contain 1 to 2 million combinatorial chemistry structures, and tens of thousands of natural product extracts. Therefore, the screening libraries are large and diverse, seeking to cover a large amount of the available chemical space.
• On average, in terms of screening numbers, one sample per well is screened, one well per sample at one concentration (of the compound), and about 400,000 samples processed per screen.
• Our conversation with one pharmaceutical company's director of HTS stated that they had performed 25 screens in 2002, with a total number of wells processed at 9.9 million.
• From a number of different participants from pharmaceutical companies, as well as the large drug discovering biotech companies, we heard that screening today is resulting in data worth up to 200 384-well plates per day for biochemical assays, and up to 130 384-well plates per day of fluorometric imaging plate reader (FLIPR) assays (a gold standard for calcium flux measurements and screening, from Molecular Devices [Sunnyvale, CA, USA]).
  • In this vein is the size of typical data sets. Up to 200 384-well plates per day, 16 compounds per plate, results in 3200 at concentrations that inhibited responses by 50% (IC50s) per day. This set of metrics frames the magnitude of the screening ensemble and market opportunity within the pharmaceutical industry.
• According to our conversations with the pharmaceutical company HTS directors, the hit rates from the primary screening process vary significantly (depending on the assay methodology, the biology being deployed, the target being studied, and the format of the screen, among other factors). The range is 0.2%-16%. Subsequently, computational algorithms help to select hits for subsequent confirmatory screening.
• On average, each screen yields about 1300 confirmed hits, according to the HTS directors in the trenches performing these screening operations within the pharmaceutical industry.
• In addition, screens for given targets are developed both by the individual therapeutic areas and by the core screening facilities within the pharmaceutical industry's operations. Also, the screens that are developed are both biochemical (in vitro) screens as well as cell-based (in vivo) screens.
• The trend within pharmaceutical companies is to migrate away from heterogeneous assay formats and toward exclusively homogeneous (mix and read) formats. Many of the large pharmaceutical companies that we spoke to at the conference said that they have already made the switch over to 100% homogeneous assay format.
• In terms of average time required for screen development and validation, our discussions with the pharmaceutical company HTS directors revealed that it takes on average 2.7 months to develop a cell-based assay (in a therapeutic area group of a pharmaceutical company). Other homogeneous assays developed similarly (i.e., under the same conditions, in the same location within industry) take on the order of 3.7 months. Homogeneous assays developed within core facilities (at a pharmaceutical company) take on average of 1.9 months to develop.
• Subsequent to the initial primary screen, the hits are reconfirmed, and these confirmed hits are then tested in subsequent assays (described below), for confirmation as Advanced Hits:
  • Best biological, chemical, or pharmacological profiles.
  • Evidence of stoichiometric interaction with the target protein (e.g., by nuclear magnetic resonance [NMR] spectroscopy, X-ray analysis, fluorescence perturbation, calorimetry, surface plasmon resonance [BiaCore; Piscataway, NJ, USA] methodology).
  • Exquisite selectivity (i.e., no cross-reactivity) in cell-based assays.
  • Only the best leads are selected for further characterization and development.
• One large pharmaceutical company mentioned that in 2002, 43% of compounds entering development were derived from hits in screens that they have performed.

Trends and Approaches that Are Quantitatively Affecting HTS Industry Metrics
Having framed in the previous section the quantitative metrics affecting the practice of HTS, we address in this section the trends (and drivers) that are impacting the current practice of HTS industry-wide. In this manner, we provide a framework for addressing the dynamics of the HTS marketplace-- both qualitatively as well as on a quantitative basis.

How much of a compound library needs to be screened? This is a major issue in the practice of HTS and impacts the throughput as well as the financial commitment to the screen that a pharmaceutical company makes. Screening of the entire compound library during a given screen.

• Driving factors.
• Provides comprehensive coverage of high value (and high priority) targets.
• Complete annotation for data mining purposes.
• Provides momentum to the screen.
• Restraining factors.
• Institutional barrier to entry.
• Certain assays cannot be brought to ultra high-throughput screening (uHTS).
• Effort, expense, and infrastructure (such as robots) required.
• Clogging of the hit verification pipeline (overload).
• Physical depletion of the compound library collection.
• Data overload (particularly of image or multiplexed assays).

Partial screening of libraries is a possibility (rather than going through the entire deck of compounds).

• The utility of such an approach is that it enables the rapid discovery of starting point(s) for chemistry.
• It enables the rapid identification of starting points for mining of compound libraries.
• Use for refinement or enrichment of particular compound libraries.
• Ligands that are expected to bind (per other physical measurements) can be tested in this manner.
• A quick determination if the target (genomics derived) is druggable, i.e., amenable to modulation of its activity by small molecules.

Virtual screening is a trend within the pharmaceutical screening enterprise. It involves the use of reference ligands to validate an ultra high throughput-screening assay.

• It involves the determination of a simplistic motif of known or proposed inhibitors of a particular target.
• Substructure search in a sample collection.
• Filter by self-similarity and motifs.
• Test in biochemical assays.
• Compare data with a full deck high-throughput screen (when completed).

The value of "front loading" via a simple virtual screen is that it results in an enrichment of confirmed hits later in the "full deck" high-throughput screen. In this manner, virtual screening helps to increase the odds of hit generation and hence saves time and reduces costs from the total drug discovery and development process. This is one of the factors affecting quantitatively the economics of the HTS marketplace and the industry dollars available to the vendors in the space.

Directed screening (fingerprinting). Since hits from cellbased assays are not generally useful in uHTS, the idea with directed screening is that researchers identify a representative weighted 25,000 compound ensemble (representative of the full deck of compounds in the library).

• Perform a cell-based assay.
• Annotate hits carefully in functional assays.
• Use for screening against rare cell types in complex low-throughput assays.
• Follow-up hits from the larger (compound) collection using fingerprint similarity searches.

In this manner, the screen of the 25,000 compound set, plus an additional few thousand compounds results in 50% of hits that would be generated using a screen of the full deck (the entire library of compounds in the collection).

Inhibitor enrichment. The objective here is to construct an enriched library that is skewed towards kinase inhibitors, for example. The approach is as follows:

• Solve structures (crystal structures) of kinases complexed with primitive inhibitors.
• Construct "composite" inhibitor space-filling models.
• Filter the available chemicals through the model.
• Acquire and test predicted inhibitors.
• Compare with several large-scale kinase high-throughput screens.
• Combine active sets.
• Test enriched set in additional kinase assays.
• Start chemistry and crystallography (in order to provide a starting point for creating focused libraries addressing kinases only).

In this manner, the industry can create very focused libraries skewed towards different (drug) target classes, thereby enabling very focused screening to take place. A key driver of this is the need to generate hits in a rapid and cost-effective manner.

In summary, several partial screening strategies show demonstrable value in pharmaceutical screening. Front loading based on virtual screening is now embedded in the screening ensemble in the pharmaceutical industry as a means to streamline the process of screening--a means of preserving precious compound libraries, cutting the time required to perform screening campaigns, as well as a way to cut costs associated with HTS operations.

These qualitative measures and strategies that the HTS ensemble is adopting have a quantitative impact on the number of screens performed per year, the size and scope of the screens, as well as the expenditures on the screens.

Emerging Technology Themes in the HTS Space
Until now, we have emphasized in this article the current quantitative metrics that define the HTS space, as well as the drivers that seek to focus the process of screening, and thereby restrain the uncontrolled growth of the HTS space. This is key given that the drug discovery and development community has allocated a certain fraction of their total research and development budget on drug discovery (25% of total R & D budget) and hence screening (including HTS, secondary screening, and absorption, distribution, metabolism, excretion [ADME] oxicity screening). In this section of the article, we present some trends in the technology side of the HTS ensemble discussed at the ScreenTech conference to illustrate where the field is migrating.

Growth of cell-based screening in HTS. This was a consistent message at the conference. Companies in the exhibit area of the conference were pushing cell-based assay products, whether it was instrumentation (such as Amersham Biosciences [Piscataway, NJ, USA], Atto Bioscience [Rockville, MD, USA; with their high-throughput confocal systems for subcellular analysis], among others), or reagents (such as Cytomyx Ltd. [Cambridge, UK], DiscoveRx [Fremont, CA, USA], Bionaut Pharmaceuticals [Cambridge, MA, USA], among others).

There were many presentations focused on the use of cell-based assays from the vendors and end-users (pharmaceutical and biotechnology companies). There were several presentations focused on cell-based approaches in HTS at the conference. Presentations ranged from instrumentation that enabled intact cellular visualization and screening to optical manipulation of individual cells. In addition, much emphasis was placed on the different biologies for different target classes [discussed later in this section].

Multiplexed assays in HTS. This is another area of increasing importance across the entire HTS community. Robert Zivin from Johnson & Johnson Pharmaceutical Research & Development Exploratory Technology Group (New Brunswick, NJ, USA) presented a major talk on the opening day of the conference about their evaluation of various multiplexing technologies and the need to multiplex in early to late HTS.

Virtual Arrays (Sunnyvale, CA, USA), another company championing the use of multiplexing to increase information output in screening, presented a talk whereby they demonstrated how their system of "digital bar codes" could be used to multiplex important biological assays. They demonstrated how GPCR assays could be multiplexed using their platform--in this way, many different GPCRs could be screened simultaneously in the same experiment. We believe that the use of multiplexed HTS assays will impact the quantitative metrics of the HTS industry.

Assay technologies that are gaining traction in HTS. The use of functional assays in HTS is growing. The days in which the majority of the screens were binding events (receptorligand binding, for instance) and their perturbation by small drug-like molecules are gone. The use of functional assays in HTS enables the direct identification of ligands and the biological effect of compounds derived from libraries.

Some of the key biologies that are gaining traction within the drug discovery community address GPCRs. This is a very successful druggable target class, and the biologies utilize reporter genes, the measurement of molecules, such as cyclic AMP (that are a byproduct of signaling cascades), and the use of engineered cell lines that have embedded in them detection systems enabling biological readout upon activation of the GPCR.

Ion channels are a growing target class, and hence, assay methodologies for addressing this class is growing. Companies such as Cytomyx Ltd. were presenting their portfolio of ion channel-transfected cell lines for use in screening. In addition, companies such as Molecular Devices were presenting their high-throughput patch clamp platform, IonWorks™, designed for thousands of ion channel assay points per day.

Protein redistribution, translocation in vivo. The use of green fluorescent protein (GFP) as an intracellular protein sensor is now firmly established in the HTS community. The use of GFP as a way of annotating cellular pathways in vivo is an emerging area. Given that GFP is well studied and there are lots of colors of GFP available, there is increasing utilization of GFP as an assay choice for protein redistribution (localization) in vivo, protein-protein interactions, protease activity, calcium measurements, and kinase activity, among others.

In summary, in this article we present the current state of the HTS landscape (circa, April 2003) from interviews and discussions with the various industry participants at recently held industry conferences (the ScreenTech World Summit). In this manner, we map out in this article the relevant qualitative and quantitative metrics in the HTS space and provide a snapshot of the technology trends that are gaining traction in this space. In the next article on this topic, we present from the ScreenTech conference, technologies that are making an impact in the screening of various different target classes.