COMBINATORIAL CHEMISTRY.

NOT JUST FOR PHARMACEUTICALS

John K. Borchardt

New synthesis techniques are increasingly used in materials science and in developing industrial chemicals, pigments, Polymers, and catalyst.

Combinatorial chemistry is the synthesis of large numbers of compounds (a library) that are systematic variants of a chemical structure (Figure 1). According to Stephen Wilson of New York University, "In the past few years, combinatorial chemistry has become the popular and often misunderstood 'new wave' in drug discovery." Its roots can be traced back to Bruce Merrifield's 1963 report V. Am. Chem. Soc., 1963, 85, p. 2149) on solid-phase peptide synthesis, for which he won the 1984 Nobel Prize in chemistry. Since then, combinatorial chemistry has expanded from peptides to organic, organometallic, inorganic, and polymer chemistry as its industrial applications have spread from the pharmaceutical industry to other industries such as catalysts, polymers, advanced materials, pigments, and agricultural chemicals. Books and journals devoted to the subject of combinatorial chemistry are published with growing regularity (see box, Additional Reading, p. 39).

Companies that have developed combinatorial chemistry techniques for the pharmaceutical industry are increasingly working with manufacturers of industrial chemicals, pigments, polymers, advanced materials, catalysts, and agricultural chemicals. The objective is more cost-effective development of improved chemicals for these applications. For example, Bayer AG (Leverkusen, Germany) and Symyx Technologies (Sunnyvale, CA) have formed a five-year research collaboration to develop new catalysts, polymers, and electronic materials. Symyx also has joint combinatorial chemistry R&D agreements with Celanese Company (Dallas, TX) for catalyst development and Ciba Specialty Chemicals (Basel, Switzerland) for the development of new pigments for plastics, coatings, fabrics, and other products. DuPont (Wilmington, DE) and Exxon Chemicals (Houston, TX) are developing combinatorial expertise on their own.

MISMATCH NOTED

The rapid exploitation of recently developed or improved technologies is at the core of the fast-paced development' of combinatorial chemistry and its use by the pharmaceutical industry. David Nickell of the Parke-Davis Pharmaceutical Research Division of the Warner-Lambert Company (Ann Arbor, MI) has noted that the introduction of high throughput screening (HTS) methods in the pharmaceutical industry in the early 1990s produced a fundamental mismatch: "Medicinal chemists were not able to generate compounds at a pace that could utilize the full capacity of HTS." Molecular modeling leads to new hypotheses of the mode of drug action and the identification of key chemical structure features that drug candidates must have. Combinatorial chemistry synthetic methods allow the rapid synthesis of a large number of candidate molecules with the desired chemical structure features (or scaffold) as well as a variety of other ancillary chemical structure features. This ability permits the insights of molecular modeling and the productivity of HTS to be better used. In fact, Cait Murray of Chemical Design (Chipping Norton, UK) says, "Me rapid progress in combinatorial chemistry now means that the number of molecules that can be made and tested far exceeds the capacity of most HTS screens." This is probably true in other industries that are beginning to use combinatorial chemistry methods.

Two different combinatorial strategies may be used to generate a large number of compounds for HTS. 'Me first, called the "split" or "pool-and-split" method, is to simultaneously create the compounds and then screen the mixture for performance. Reactions occur in the solid phase; compounds are, attached to polymer beads. After a reaction, the beads are split into separate containers, and different agents are added. 'Me beads containing the products of these reactions are mixed (or pooled), then split again and reacted with different reagents. The process is repeated a certain number of times. Tags attached to the beads at each step indicate the reagents with which each bead has been treated. After the synthesis is complete, the beads are screened for performance, the chemical structure of the most active compound in the mixture is identified, and this compound is prepared using standard synthetic methods.1his combinatorial method is best suited for the synthesis of many (up to one million) compounds called a library. "The attraction of the combinatorial approach is the speed and cost -effectiveness of generating molecular diversity- the challenge is to develop methods to readily identify the active component in a complex mixture either through labeling or statistical methods," notes G. Paul Savage, project leader of the Crop Projection Chemicals Project for Dunlena, an Australian joint venture company of DuPont and the Commonwealth Scientific and Industrial Research Organization.

The other combinatorial method is the parallel synthesis approach. All chemical structure combinations are prepared separately, in parallel, on a given chemical structure "scaffold" using an automated robotic synthesis apparatus. Hundreds or even thousands of vials are used to perform these reactions, and laboratory robots are programmed to deliver specific reagents to each vial. Although this approach is automated, it often takes longer than the split synthesis method to complete and thus is best suited for the development of smaller chemical libraries.

CATALYST DEVELOPMENT

The initial mechanism of drug action is the specific binding or docking of a portion of the drug candidate to a receptor site. 'Ibis action is conceptually similar to that of many catalysts in promoting chemical reactions, particularly polymerizations. So, it is not surprising that combinatorial chemistry is being used to synthesize catalyst candidates. Symyx is developing industrial relationships with several chemical companies in which Symyx designs and performs candidate catalyst screening projects.

Ligands attached to metal ions play an important role in catalysis. Burger and Still reported the reaction of tetraazacyclodecane units with three functionalized side chains with other chemicals using combinatorial techniques to create a large number of ligands (1). These were then reacted with divalent cobalt and copper ions to synthesize many catalyst candidates on polymer beads. Combinatorial techniques have also been applied to the synthesis of bipyridyl and crown ether ligands (2). Metal template-directed complexation has been used to prepare a small combinatorial library of catalyst candidates in solution.

Menger et al. synthesized libraries of poly(allylamine)-based catalysts by reaction of the polymer with mixtures of functionalized carboxylic acids (3). The polymeric mixture was treated with various metal salts and then examined for phosphatase catalytic activity in hydrolyzing bis-(p-nitrophenyl)phosphate. With the use of sub-library synthesis, the best catalyst systems could be determined. Similar techniques could be used to develop catalysts for industrial applications.

At the 1998 ACS fall national meeting in Boston, Pennsylvania State University researchers reported using combinatorial techniques to vary the surface structure of nanoparticulate titanium dioxide doped with noble metals. The quantity and composition of the dopant was varied, and the library of catalyst candidates was characterized and evaluated in terms of catalyst efficiency and activity.

Also at that meeting, a University of Houston team reported developing a combinatorial library of candidates of catalysts for oxidation reactions. Each catalyst consisted of a different metal/ promotor loading on a single pellet. Noninvasive infrared thermography was used to measure the heat liberated in the catalytic reaction and thus evaluate catalyst activity.

In developing new catalysts, measuring reaction rates is often difficult. At the University of North Carolina, James Morken and graduate student Steven Taylor correlated the reaction rates of a library of 3,150 catalyst candidates supported on resin beads with the amount of heat liberated to the solvent during the reactions. Morken explains, "Police departments use infrared devices to pick up beat produced by children lost in the woods. Similarly, we use a thermographic video camera to find which polymer beads -to which chemicals have been attached- are generating heat. The heat indicates a chemical reaction, and the more heat generally the better that chemical is as a catalyst." The most active catalysts were those most strongly bound to the beads. Morken and Taylor used the eye of a needle as a ladle to scoop out the hottest beads (i.e., the best catalyst candidates) from the single reactor in which.the reaction occurs simultaneously over the combinatorial library of catalyst candidates. Using this technique, the chemical structures of only the best catalyst candidates need be determined, thus saving significant amounts of time.

"With combinatorial chemistry," Morken notes, "you can make 100,000 compounds in about two weeks. It should not take our assay a whole lot longer than that to screen those. We believe this work is going to have a large impact on the catalyst discovery process." A former colleague of Morken's, Nathaniel Finney (University of Califonnia-San Diego), says, "This is the most significant advance yet reported in the application of combinatorial methods to the discovery of new highly active catalysts. While the synthesis of polymer-bound catalysts has a long history, the simultaneous evaluation of mixtures of catalysts has remained an elusive goal. Morken and Taylor have provided an elegant clever solution to this screening problem, and one which promises to be extremely general."

Zeolites have been used as catalysts in oil refineries and in other chemical plants, as ingredients in laundry detergents, and in various other applications. Norwegian researchers have developed an autoclave that can synthesize at least 100 zeolites simultaneously at temperatures up to 200 'C (4). This development extends the capabilities of combinatorial synthetic techniques to harsher reaction conditions than previously demonstrated.

MATERIALS DEVELOPMENT

Combinatorial chemistry is increasingly being applied to solve problems in materials science. For example, as electronic devices get smaller, some electrical properties (e.g., capacitance change, improved thin-film insulators) are needed. Using combinatorial chemistry, a Bell Labs group (Lucent Technologies, Murray Hill, NJ) has developed a thin-film material that has a higher dielectric constant than silicon dioxide, the insulator most commonly used in dynamic random-access memory (DRAM) computer chips. To develop their optimum thin-film insulator candidate, a mixed oxide of zirconium. tin, and titanium (Zro.15Sno.3Ti 0.5502)9 they created more than thirty combinatorial libraries containing about 4,000 compounds each. A team from the University of California-Berkeley (UCB) and Lawrence Berkeley National Laboratory (LBNL; Berkeley, CA) also used combinatorial techniques to synthesize a barium strontium titanate system for use in microwave devices.

Over many years and despite intensive research, fewer than 100 useful commercial phosphor materials have been discovered through conventional one-by-one synthesis and testing. Phosphor materials are of great importance for developing improved flat-panel displays and lighting. The UCB-LBNL group has used combinatorial. techniques combined with analysis using a parallel imaging system and a scanning spectrophotometer to synthesize and identify an efficient blue photoluminescent material, a mixture of gadolinium gallium oxide (Gd3Ga5Olo and silicon dioxide (SiO2).1he luminescence is believed to result from interfacial effects between the two compounds. Another combinatorial method has been used to produce a library of more than 25,000 thin-film candidate phosphors and identify a new red phosphor, a vanadate of yttrium, aluminum, lanthanum, and europiurn (Y0.845AI0.070L.a0.060Eu0.025VO4), that has a quantum efficiency equal to or greater than those of existing commercial red phosphors (5).

Photoreactive materials show promise for applications such as optical data storage and image amplification. However, polymer crystallization in these materials is often a problem. A trio of University of California-San Diego researchers reported at the 1998 ACS fall national meeting that they used combinatorial chemistry to prepare a library of materials with different optical and physical properties and identify a candidate that could overcome this crystallization problern.1be parallel synthesis is based on grafting components to siloxane polymers by hydrosilation and postgraft chemical synthesis.

OVERCOMING PROBLEMS

The UCB-LBNL group has applied cornbinatorial chemistry to the problem of developing improved hightemperature superconductors by depositing combinatorial arrays of inorganic salts to create thin films of potentially superconducting composites. Each film occupies a space of 200 x 200 jmicrom in an array that can accommodate 10,000 different materials per square inch. These candidates were evaluated for high-temperature superconductivity using a specially designed robotic microprobe. This research group also has applied the combinatorial chemistry technique to magnetoresistive materials.

One limitation of the use of combinatorial chemistry in industrial chemistry is that contamination can obscure results obtained with volatile and reactive chemicals of limited stability (e.g., some organometallic compounds). Improved synthesis apparatuses can help overcome this problem. Several companies are developing improved combinatorial synthesis, analysis, and screening apparatuses. For example, a specialized robotic fluid handler that can handle air- and moisture- sensitive reagents is the key feature of an automated parallel synthesizer. Capable of handling systems such as moisturesensitive trimethylaluminum in toluene, the synthesizer extends the range of chemistries that can be used in combinatorial synthesis. Sibia Neurosciences (La Jolla, CA) has developed a method to create and preserve reactive intermediates on resin beads. Rapp Polymere (Ttibingen, Germany) and Argonaut Technologies, Inc. (San Carlos, CA), have developed improved resins and resin-bead technology for solid-phase synthesis.

The combinatorial reactor designed to synthesize zeolites at temperatures up to 200 'C has already been mentioned. This development indicates it should be possible to design combinatorial synthesis apparatus capable of performing high-temperature and high-pressure reactions.

Laboratory robotics manufacturer Zymark Corp. (Hopkinton, MA) has jointly developed with Zeneca Pharmaceuticals (Wilmington, DE) a multiple-parallel solid-phase synthesis apparatus driven by highly flexible software. Once a promising candidate compound has been developed, large quantities of the compound are needed for product testing and development. Process conditions need to be optimized to define the most cost-effective commercial manufacture of the candidate. Bohdan Automation (Mundelein, IQ and Gilson (Middletown, WI) have cooperatively developed a workstation designed for automated process development that allows variation of many reaction parameters such as time, temperature, and pH to optimize reaction rate and yield. The workstation can perform twelve reactions simultaneously.

Trega Biosciences (San Diego, CA) has developed a centrifuge that features convenient decanting of liquids from solids. IRORI (La Jolla, CA) has created tagging methodologies for both small and large compound libraries. For larger libraries, this device can be used for the unattended sorting of up to 10000 microreactors. The reactors are put into forty-eight wells for pool-and-split synthetic protocols. Gilson, Argonaut Technologies, and Biotage (Charlottesville, VA) have developed improved apparatuses for preparation and analysis of combinatorial chemistry libraries. Panlabs (Seattle, WA) uses the Biotage Parallex (which contains Gilson components) to characterize the products of 8,000 cornbinatorial chemistry reactions per month.

Another possible limitation on the use of combinatorial chemistry is the challenge of analyzing the complex compound mixtures produced in combinatorial synthesis. Mass spectrometry is an excellent tool for analyzing these mixtures, partly because of the speed of the technique. However, MS alone is inadequate when compounds in the combinatorial library include stereoisomers or positional isomers or happen to have the same molecular weight. The combination of HPLC and the use of MS/MS techniques can overcome this problem, but with significantly longer analysis time.'Ibe recent development of HPLC-nuclear magnetic resonance (NMR) techniques permits the analysis of individual components in complex mixtures. Researchers at Novartis Pharmaceuticals Corporation (East Hanover, NJ) have developed on-flow NMR techniques that can be used with the technique to decrease analysis time. A robotic liquid sample handler has been integrated with a capillary gas chromatograph fitted with an electron capture detector to improve the efficiency of gas chromatographic analysis.

THEFUTURE

Use of combinatorial chemistry in pharmaceutical R&D significantly predates its use in other industries. However, without the need for the lengthy intensive testing associated with new drug development, industrial chemicals developed using combinatorial methods could beat many combinatorial drugs to market. (Toxicological and environmental testing of industrial chemicals is required; however, this testing is far less time-consuming than the clinical testing required for the commercialization of a new drug.)

The examples of combinatorial chemistry cited in this article constitute a list that is far from complete. However, they indicate the many different applications of combinatorial chemistry in the chemical industry. Although published work still centers on solid-support synthesis-such as the catalysts on polymer beads synthesized and evaluated by Taylor and Morken-work on solution-phase combinatorial chemistry is increasing. Because of the vast number of candidates becoming available for testing, new challenges have arisen in several areas. These challenges include how to store and use combinatorial libraries of chemicals and especially how to intelligently select subsets of chemicals for testing. For many applications, improved methods of HTS are needed for the many compounds that need to be evaluated. Finally, companies are working on better methods of storing, organizing, and using the large volumes of data generated by combinatorial methods. New developments in these areas will be reported at the "Combinatorial Approaches for New Materials Discovery" conference, 21-22 January 1999 in San Jose, CA (wwwknowledgefoundation.com).

The cost-effectiveness of combinatorial chemistry technology suggests that the technology will increasingly be used in the design of many chemicals used in industrial processes and found in consumer and industrial products. As a result, many industrial research chemists, engineers, and laboratory technicians in many different fields will have to become familiar with the techniques and practice of combinatorial chemistry.

References