Random screening, while seemingly wasteful, has an important place in developing lead compounds in areas in which theory lags. Screening for antitumor activity has been carried on for more than 30 years by the U.S. National Cancer Institute, with tens of thousands of compounds being tested on tumors in vivo and in vitro. More recently, a computerized prescreening method has been applied to this process, saving time and expense, and hence the screening is not as random as it used to be. A successful random search for antibacterial action was conducted by several pharmaceutical companies in the 1950s. They tested soil samples from all over the world, which resulted in the dis-covery of many novel structures and some spectacularly useful groups of antibiotics, notably the tetracyclines (3.4). In fact, microbial sources have supplied an enormous number of new drug prototypes, sometimes of staggering complexity. Recently, the large-scale automated testing of microbial mutants has been realized and combined with recombinant DNA techniques to speed up the efficient discovery and production of new antibiotics.
Some would argue that drug discovery through screening provides the “irrational”
counterpart to rational drug design. This remark is unjustifiably harsh and is somewhat
facetious. As mentioned, screening of compounds has a long and rather illustrious history and has produced many useful anticancer and antibiotic drugs. The discovery of the anticonvulsant drug phenytoin provides an early example of drug discovery through screening.
3.2.6.1 Drug Discovery by Screening: Diphenylhydantoin, An Early Example As a chemical species, the hydantoins have been known since the 1860s. By the latter half of the nineteenth century numerous hydantoin analogs had been synthesized, but only one, 5-ethyl-5-phenylhydantoin (nirvanol), demonstrated any clinical utility.
Wernecke introduced nirvanol in 1916 as a “less toxic hypnotic”; however, enthusiasm rapidly waned when its chronic toxicity became recognized. Not surprisingly, a second hydantoin, 5,5-diphenylhydantoin (phenytoin), which had long remained on the labora-tory shelf, appeared doomed to obscurity; phenytoin had been first synthesized by Biltz in 1908, through a condensation of urea with benzil which exploited a pinacolone rearrangement.
In the late 1930s, T. Putnam initiated a screening programme to search for new anti-convulsants, using protection against electroshock-induced convulsions as a selection criterion. A makeshift apparatus to execute these experiments was assembled using a commutator salvaged from a World War I German aircraft. Having studied the structure of phenobarbital, Putnam randomly requested a diverse selection of heterocyclic phenyl-containing compounds from a variety of chemical manufacturers. He also com-municated with a number of pharmaceutical companies. The Parke-Davis Company provided nineteen heterocyclic phenyl-substituted compounds that had been deemed
“worthless hypnotics.” Phenytoin was one of these nineteen compounds. Putnam screened hundreds of compounds but only phenytoin combined high activity with low toxicity. In 1936, Putnam’s colleague, Houston Merritt, initiated a clinical evaluation of phenytoin, which soon led to its widespread marketing as an anticonvulsant drug.
The pioneering screening techniques that heralded the discovery of phenytoin pro-foundly influenced subsequent antiepileptic drug discovery. Hundreds of hydantoin analogs were synthesized and screened for biological activity; hundreds of other penta-atomic heterocyclic compounds (e.g., succinimides, oxazolidinediones) were likewise synthesized and screened for biological activity. Many of these new compounds found their way into the market place, with varying degrees of therapeutic success.
3.2.6.2 Drug Discovery by Screening: A Modern Definition
Thankfully, the science of drug discovery by screening has advanced since the time of Merritt and Putnam. Modern drug discovery by screening is more of a systematic tech-nological tour de force than a hit-or-miss gamble. The reasons for these advances are obvious. Although rational drug design is elegant, it is also slow and thus time-inefficient.
It takes a long time to identify the proteins that are involved in a disease, then crystallize them and design drugs to bind to them. Worse, some proteins, especially membrane- bound proteins, seem to defy crystallization, while some diseases do not even have identifiable proteins involved in their pathogenesis and etiology. Screening methods attempt to address all of these deficiencies in drug discovery.
If a reliable bioassay is available, it is possible to screen thousands or even millions of compounds against this bioassay. The crystal structures of key protein receptors do not have to be known; indeed, the proteins do not even have to be identified. If the bioassay is fast and efficient, and if the library of compounds being screened is diverse and comprehensive, then in principle it should be possible to identify a lead compound years before the practitioner of rational drug design. However, the key to success lies in the “goodness of the library of compounds” (i.e., a combinatorial chemistry library) and in the “goodness of the screening bioassay” (i.e., high throughput screening methods).
3.2.6.3 Combinatorial Chemistry and Drug Discovery by Screening
A key to success in drug discovery by screening is the availability of a large and struc-turally diverse library of compounds. If the library contains a million compounds that are all analogs of each other, then it may be large but it is probably not sufficiently diverse. The library should have the full range of functional groups (cations, anions, hydrogen bond donors, hydrogen bond acceptors, lipophilic, aromatic, etc.) displayed in all possible permutations and combinations in three-dimensional space. Creating such a library is not a trivial task.
Combinatorial chemistry is both the philosophical and the practical method with which to create structurally diverse compound libraries. Combinatorial chemistry is defined as that branch of synthetic organic chemistry that enables the concomitant syn-thesis of large numbers of chemical variants in such a manner as to permit their evalu-ation, isolevalu-ation, and identification. Combinatorial chemistry affords techniques for the systematic creation of large but structurally diverse libraries. From a technical perspec-tive, there are several avenues of approach to library creation:
1. Libraries of oligomers of naturally occurring monomers a. Oligopeptide libraries
b. Oligonucleotide libraries
2. Libraries of oligomers of non-naturally occurring monomers a. Oligocarbamate libraries
b. Oligourea libraries c. Oligosulfone libraries d. Oligosulfoxide libraries
3. Libraries of monomers with multiple sites for substituents a. Synthetic ease privileged structure
Historically, the first major libraries were oligomers of naturally occurring monomers.
A good example would be a library of all possible tripeptides. Using the twenty naturally
occurring amino acids, it is possible to produce 8000 different tripeptides. If atypical amino acids and amino acids in the unnatural D configuration are included, it is possible to achieve 125,000 different compounds with relative ease. Peptide libraries are easy to synthesize and, since amino acid side chains possess a wide variety of different func-tional groups, it is possible to achieve a good measure of structural diversity. However, in general, peptides are not drugs and a peptide lead would have to be modified into a drug-like molecule. In addition to oligopeptides, other naturally occurring oligomeric libraries are possible, including oligonucleotide libraries.
A step beyond the naturally occurring oligomeric libraries is to create libraries from non-naturally occurring monomeric building blocks. The medicinal chemistry literature contains a fair number of examples of such libraries, including oligocarbamates and oligoureas. Although these libraries overcome the limitations of naturally occurring oligomeric libraries, most drugs are not polymers.
To address this problem, new libraries emerged in which the central moiety was a small organic molecule. The diversity library was then constructed by attaching many different substituents to this central moiety. Some of these moieties were selected because they were very simple to synthesize. For example, dioxapiperazines are cyclic dipeptides and thus are relatively trivial to prepare. Other monomers were selected because they had a good track record for being drug-like molecules. Benzodiazepines are a good example of such libraries.
In preparing these various libraries, extensive use is made of solid phase synthetic methods. These methods are all derived from the solid phase peptide synthesis (SPPS) method developed by Merrifield in 1963. When performing a large number of synthe-ses, it is preferable to perform the synthetic steps on a solid bead rather than complet-ing the entire synthesis in the solution phase. The solid-phase technique makes byproduct removal and final compound purification easier. The organic chemistry literature con-tains a wealth of different types of solid-phase supports and novel linkers for attaching the synthetic substrate to the bead.
3.2.6.4 High Throughput Screening and Drug Discovery
If a large, chemically diverse library is available, the next problem is to evaluate these compounds in a time-efficient manner. If a 200,000 compound library is available, the biological evaluation assay must be rapid and reliable. If the assay were capable of test-ing five compounds per day, it would take 110 years to evaluate the entire library.
Clearly, this is not the time for elaborate in vivo testing. Fast, efficient in vitro assays are required. The ability to inhibit an enzyme is a good example of a potentially useful assay for high throughput screening.
A variety of high throughput assays have been developed and perfected over the past 10–20 years. These include the following basic types of assay:
1. Microplate activity assays (assay is in solution in a well; the result of the assay, such as enzyme inhibition, is linked to some observable, such as color change, to enable identification of bioactivity)
2. Gel diffusion assays (biological target is mixed in soft agar and spread as a thin film;
the compound library is spread on the surface of the film; after allowing for compound
diffusion, an appropriate developing agent is sprayed on the agar surface and areas in which bioactivity has occurred will show up as distinct zones)
3. Affinity selection assays (compound library is applied to a protein target receptor; all compounds that do not bind are removed; compounds that do bind are then identified) Of these, microplate assays are probably the most widely used. Screening combinator-ial libraries in 96- or even 384-well microplates is time and cost efficient. Using modern robotic techniques, it is possible to perform more than 100,000 bioassays per week in a microplate system (permitting the above-described 200,000 compound library to be screened in two weeks, rather than over a century).
In addition to selecting an appropriate assay, it is also necessary to have a pooling strategy. It is more efficient to test many compounds per well on the microplate, rather than one. If one could test 100 compounds per well, then the standard 96-well plate would enable almost 10,000 compounds to be evaluated in one experiment. (Currently, multiwell plates containing more than 96 wells are routinely being used.) To facilitate effective pooling, the library of compounds is usually divided into a number of nonoverlapping subsets.
The synthetic strategy employed during the combinatorial syntheses can be used to assist in determining these pooling strategies. In random incorporation syntheses, a single bead could contain millions of different molecular species. In mix and split syn-theses (also called pool and divide synsyn-theses or one bead–one compound synsyn-theses) only one compound is attached to any given solid-phase synthetic bead.
The evolution of methods for combinatorial syntheses and high throughput screening will be necessary to address the explosion of druggable targets soon to be identified by the genomics and proteomics revolutions. Genomics and proteomics represent the future of lead compound identification.
3.2.7 Lead Compound Identification through Genomics and Proteomics