Combinatorial Chemistry

Chemical & Engineering News
April 6, 1998
Copyright © 1998 by the American Chemical Society

Industry is embracing the technology 'totally,' as researchers continue to advance the art of rapid synthesis and screening Stu Borman C&EN Washington Combinatorial chemistry-a technology for creating molecules en masse and testing them rapidly for desirable properties-continues to branch out rapidly. Compared with conventional one-molecule-at-a-time discovery strategies, many researchers see combinatorial chemistry as a better way to discover new drugs, catalysts, and materials. Combinatorial chemistry's potential for accelerating drug discovery-its primary focus so far-has proven to be so compelling that few if any pharmaceutical companies have been able to ignore it, even if they may have wanted to. And many new companies have sprung up in an effort to reduce combinatorial chemistry to practice.
Houghten: everyone jumped on this	"If you go back 10 years there was really nothing out there, and then all of a sudden, starting about three years ago, everyone jumped on this," says Richard A. Houghten, president of Torrey Pines Institute for Molecular Studies, San Diego, and one of the pioneers of the field." It's really quite spectacular how quickly it's become almost an accepted routine. Whereas 10 years ago a good medicinal chemist might make 50 to 100 compounds a year, that same chemist is probably expected to make in the thousands or tens of thousands today."
Myers: it's doing an awful lot of good	Peter L. Myers, chief scientific officer of CombiChem, San Diego, agrees. "The pharmaceutical industry, in particular, has embraced it totally. Every company now has some aspect of this in-house."
The achievements of and prospects for combinatorial chemistry were the main focus of "Exploiting Molecular Diversity" and "High-Throughput Organic Synthesis"-back-to-back conferences sponsored by Cambridge Healthtech Institute, Newton Upper Falls, Mass., and held last month in sunny Coronado, Calif. The topics discussed there were wide-ranging: business developments; the use of biological organisms and enzymes to construct collections of compounds (known as libraries); combinatorial information management (a topic C&EN plans to cover in its April 27 issue); miniaturized systems for synthesis and screening; strategies and methods for combinatorial synthesis; automated instrumentation; and applications. At the meetings, Houghten pointed out that the origins of combinatorial chemistry can be traced back at least as far as 1963, when biochemistry professor R. Bruce Merrifield of Rockefeller University, New York City, developed a way to make peptides by solid-phase synthesis. But the field in its modern dimensions only began to take shape in the 1980s, when research scientist H. Mario Geysen, now at Glaxo Wellcome, Research Triangle Park, N.C., developed a technique to synthesize arrays of peptides on pin-shaped solid supports and Houghten developed a technique for creating peptide libraries in tiny mesh "tea bags" by solid-phase parallel synthesis. Although peptide and oligonucleotide libraries are still very much of interest, the combinatorial field has gravitated toward libraries of druglike small organic molecules. In the early 1990s, the groups of chemistry professor Jonathan A. Ellman of the University of California, Berkeley, and organic chemist Sheila H. DeWitt, then at Parke-Davis Pharmaceutical Research, Ann Arbor, Mich. (now director of business development at Orchid Biocomputer, Princeton, N.J.), independently worked out some of the earliest techniques for producing small-molecule libraries.
Orchid Biocomputer's 2-inch square microchips carry out submicroliter chemical reactions in a 144-well array format.	A passing fad? With combinatorial chemistry now having been up and running for several years, C&EN asked conference participants if combinatorial chemistry is indeed a viable methodology for drug and materials discovery or just an overhyped fad. From the responses, "combichem" clearly is more than just the latest hot trend.
"You're seeing reality setting in and the original euphoria fizzing out a little bit," said Myers in regard to people's perceptions of combinatorial chemistry. "Structure-based drug design went through that sort of evolution." He added, however, that he believes combinatorial chemistry "actually is doing an awful lot of good-it's just not too visible yet. Unfortunately, people don't publish at a very early stage. Eli Lilly has a combinatorial drug in development, and many other companies have compounds in Phase I clinical trials, based on lead optimization using library synthesis." (Leads are compounds that show promising activity and are potentially capable of being developed into drugs.) "It's not so easy to say whether any company has yet discovered a series of fundamentally different structures that could not have come from traditional sources by screening combinatorial libraries for lead generation," Myers continued. "But I have no doubt that for lead optimization, combinatorial chemistry is working and is a major contributor to analog synthesis."
Czarnik: enormous investments made	"Isn't the record to date pretty poor for all the effort that's been devoted to this field?" Anthony W. Czarnik, vice president for chemistry at Irori Quantum Microchemistry, La Jolla, Calif., asked rhetorically. "Even if this were true, the same could be said of the whole drug-discovery process. While the record is good, the cost per drug is historically high. Chemists don't engage in drug discovery because it's fast, but rather because it's important and valuable."
If combinatorial methodologies were not perceived as valuable to the drug-discovery process, said Czarnik, "companies would not have made the truly enormous investments they already have. And I've been pleasantly surprised to learn that these methods are continuing to migrate into more mainstream medicinal chemistry labs. A few years ago, this was not so clear." Houghten added: "Many skeptics ask, 'If combinatorial chemistry techniques are so good, why aren't more new drugs being generated by these methods?' That's a legitimate question, but you have to realize that once you have identified a lead compound, it takes three to 10 years to work it through the clinical trials process and get Food & Drug Administration approval." In addition to Eli Lilly's combinatorial drug, said Houghten, "Trega Biosciences [San Diego] has a compound in clinical trials for pain and asthma that came out of a combinatorial chemistry program; Magainin Pharmaceuticals [Plymouth Meeting, Pa.] has a compound in trials that was clearly based on early combinatorial approaches; and it's my understanding that Merck has a combinatorial compound in clinical trials." However, said Houghten, "the speed with which these methods have been accepted and incorporated into routine discovery practices at large pharmaceutical companies makes it increasingly difficult to determine which new drugs were found using combinatorial chemistry." A case in point is the 1992 discovery of an orally active thrombin inhibitor by Thomas R. Webb, then at Corvas International, San Diego, and now director of chemical services and business development at ChemBridge, also in San Diego. Webb constructed one of the earliest libraries of small organic molecules by solid-phase parallel synthesis and found the thrombin inhibitor in that library. The inhibitor entered clinical trials, but few are aware that it was identified combinatorially. Journals One indication of the vibrancy and staying power of a field is its coverage by specialized scientific journals, and the number of such publications in the combinatorial chemistry field is on a decided upswing. Molecular Diversity, which has six founding coeditors and is put out by Kluwer Academic Publishers, Dordrecht, the Netherlands, is the pioneer combinatorial journal, having covered the field since 1995. It publishes basic and applied papers dealing with combinatorial chemistry, screening of library compounds against various targets, and applications. But three additional journals are about to make their debuts. The first issue of Combinatorial Chemistry & High Throughput Screening, published by Bentham Science Publishers, Hilversum, the Netherlands, is due to appear this month. The journal's editor-in-chief is Richard B. van Breemen of the department of medicinal chemistry and pharmacognosy of the University of Illinois, Chicago. It will cover a range of topics in combinatorial chemistry (including small-molecule, peptide, nucleic acid, and phage-display libraries) and high-throughput screening. Papers on related topics, such as robotics and informatics, will also be included. John Wiley & Sons, New York City, is introducing Combinatorial Chemistry,with the first issue also due out this month. The editor is David J. Ecker, vice president and managing director of combinatorial drug discovery at Isis Pharmaceuticals, Carlsbad, Calif. "A lot of work that's done in the field has nowhere to go because of the interdisciplinary nature of combinatorial chemistry," said Ecker. "It may not be organic enough for the Journal of Organic Chemistry, it may not be sufficiently automation-related for an instrumentation journal, and it may not be pharmacological enough for a pharmacology journal. From a survey of the community, I got considerable sentiment that there was a need for a workhorse, nuts-and-bolts publication that covered the real work of people in the field- the kind of journal you would see at the end of a lab bench where libraries were being made." Combinatorial Chemistrywill be such a journal, he said. In addition, the American Chemical Society has just approved publication of the Journal of Combinatorial Chemistry,with the first issue scheduled to appear in January 1999. Czarnik, editor of the journal, said: "There's an interesting issue about what level of characterization should be required for work to be publishable. If you make 1,000 compounds in a library, the reality is that you can't get an NMR spectrum on every one of them, so you can't meet the publication standards of, say, the Journal of Organic Chemistry." On the other hand, Czarnik said, "a lot of combinatorial papers are published without any characterization of the compounds and even without a description of the experimental method, so people can't repeat the work easily. I think this new journal has a good chance of addressing these problems by setting publication guidelines for the field, which up to now has just been like the Wild West."
Persidis: combi power is impressive	Combi power Aris Persidis, vice president of business development at Argonex, Charlottesville, Va., is among those who believe that combinatorial chemistry is very much on the upswing. "The reason we're here today," he said on the first day of the Cambridge Healthtech Institute meetings, "is combi power. Using traditional organic chemistry takes a chemist one month to produce four leads, at a cost per compound that is estimated to be about $7,500. In that same period, an organic chemist employing combinatorial approaches would have produced about 3,300 compounds ...at a cost per compound of about 12 bucks. I like to call that combi power, and I think we agree that that is indeed very impressive."
Combinatorial chemistry has been referred to in various ways, such as "molecular evolution," "in vitro selection," and" molecular diversity." But Persidis said he believes there is a need for a term that is more encompassing. He suggested "combinatorics," reflecting the field's high-throughput screening, automation, and informatics aspects in addition to its chemistry component. "Combinatorics also has the requisite' ics' ending," said Persidis. "Everything that attracts a lot of money these days has an 'ics' ending"-including genomics (the development of new gene targets) and proteomics (the study of new protein targets), two disciplines that are increasingly being melded together with combichem in drug-development efforts. "There are about 1,800 to 1,900 biotech companies in the world, and right now about one-tenth of them have a formal program in combinatorial chemistry," said Persidis. A broad network of combinatorial alliances, consortia, contracts, and agreements also has been established, with a total value of about $2.6 billion. Overall, "the market capitalization of combi-specific companies has risen 100-fold since 1990, and the capital they have raised has risen 65-fold," said Persidis, quoting data from the Biovista "Combinatorics" report, published by biotechnology research firm Assett, Athens, Greece. "The acid test of validation in any platform industry in biotech is merger and acquisition activity," Persidis said, and combinatorial chemistry has been showing its mettle in this area as well. As examples, he noted the 1994 purchase of Sphinx Pharmaceuticals, Durham, N.C., by Eli Lilly for about $80 million; the acquisition of Selectide, Tucson, Ariz., by Marion Merrell Dow a year later for $58 million; and what he called "the biggie"-the 1995 purchase of Affymax Research Institute, Palo Alto, Calif., by Glaxo Wellcome for $533 million. Combi-consortia that have been formed include the Diversity Biotechnology Consortium put together in 1994 by the Santa Fe Institute (in New Mexico); the Combinatorial Chemistry Consortium created in 1996 by Molecular Simulations, San Diego; and one that currently is being organized by Systems Integration Drug Discovery Co. (SIDDCO), Tucson. According to Colin Dalton, SIDDCO cofounder and executive vice president, the consortium model is a cost-effective way for companies to gain access to the personnel and information required to make libraries. Partners in SIDDCO's consortium will have shared access to synthesis development services, automated instrumentation, and a reaction database, but will still be able to maintain the confidentiality of their proprietary libraries, screening hits, and optimized lead compounds. "SIDDCO's consortium should be appealing to product-based biotech companies that recognize the value of combichem but do not have the resources to establish their own program," said Dalton. "We have three partners signed up and expect to fill the consortium with six partners." Research scientists at EnzyMed with an instrument for automated workup of biocatalytic transformations. The field also is characterized by a complex web of one-on-one collaborative deals. For instance, Amylin Pharmaceuticals, San Diego, has a series of collaborative arrangements-with Tripos (St. Louis), EnzyMed (Iowa City, Iowa), ChemDiv (Glendale, Wis.), and SIDDCO-to screen those companies' proprietary libraries for metabolic and cardiovascular agents. Nigel R. A. Beeley, Amylin's senior director of chemistry, explained that his company doesn't have "much money to invest in acquiring large numbers of samples for screening purposes. Our medicinal chemists are really focused on optimizing hits, rather than generating large libraries of molecules to identify hits." The collaborations, he said, "offer us a way of acquiring samples so we can develop a pipeline of products." Is combichem primarily industrial? The extent of industrial combi power suggests that combinatorial chemistry is becoming primarily a commercial, as opposed to an academic, endeavor. Although important combinatorial technologies such as solid-phase synthesis have emerged from academia, said Myers, the field "now has become much more of an applied science-primarily for the development of pharmaceuticals and agrochemicals, but increasingly for foods, metals, and catalyst discovery as well." Czarnik agreed that combinatorial research is indeed now centered in the corporate sphere, but he named more than a score of academicians who have made and who continue to make major contributions to the field. To the extent that combinatorial efforts are centered in industry, said Houghten, "the reason is almost entirely economic. To make 100,000 individual compounds and screen them in a high-throughput array format is typically prohibitive for most academic organizations. If you need to look at 100,000 individual data points in triplicate, that will almost always short-circuit academic organizations. They just can't get to it." Chemistry professor Kim D. Janda of Scripps Research Institute, La Jolla, agreed with Houghten's point about economics." I cannot compete with a major pharmaceutical or biotech company to make libraries," said Janda, noting that government grants for combinatorial research can be difficult to get. "Basically, to continue in the combinatorial arena I've had to secure money from industrial companies." Daniel V. Santi, president and chairman of Kosan Biosciences, Burlingame, Calif., likewise agreed that "doing combinatorial chemistry the way it's looked at today is a very expensive business. It's very instrument-intensive, and it's difficult to do in academia. Even though Kosan's technology was invented at Stanford, it's pretty hard to hit it from an academic lab." Nevertheless, the recent creation of university-based combinatorial chemistry centers is but one sign of enduring academic confidence in the field. The University of Pittsburgh's Combinatorial Chemistry Center, directed by chemistry professor Peter Wipf, is focusing on solid-phase-synthesis strategies, purification techniques (such as fluorous synthesis and phase switching), catalyst design, and automation for natural product synthesis and process chemistry. And the new Combinatorial Center at York University, Toronto, established in collaboration with several pharmaceutical and instrumentation companies and directed by chemistry professor Michael G. Organ, specializes in solution- and solid-phase synthesis of small-molecule libraries. In terms of the employment status of chemists in general, when combinatorial chemistry first emerged concerns were expressed that the technology was so efficient that it would to some degree boot the chemist out of chemistry. But attendees at the meetings seemed to believe that just the opposite has occurred: The chemist now is more important than ever. "Parallel synthesis of novel, chemically complex, structurally diverse, and druglike small molecules is still presenting a great challenge, and the chemistry itself starts to be a key issue and major limiting factor in the field," said Eugene F. Vaisberg, president and chief executive officer of ChemBridge. "It is not surprising that despite earlier concerns that combinatorial robots would substitute for chemists in the labs, the demand for good chemists is actually going up dramatically." Senior director of lead discovery Richard F. Labaudiniere at Rhône-Poulenc Rorer, Collegeville, Pa., said he believes that increasing numbers of medicinal chemists not only are needed to optimize leads by combinatorial methods, but also to fine-tune lead activity by classical methods. Libraries Each combinatorial chemistry company seems to have its own unique strategy for generating libraries. For example, ChemBridge researchers produced the company's new PharmaCore lead generation library by combining diversity analysis, informatics, and traditional medicinal chemistry techniques in the design stage. During the actual synthesis phase, synthetic intermediates were used to construct a "synthon" collection that "makes it possible to generate follow-up libraries to rapidly establish structure-activity information," said Webb. "Obtaining such data is often the slow step in the discovery of clinical candidates." The structure-activity data can be used, he said, "to quickly prepare more active derivatives and, after several iterations, optimal compounds ready for preclinical evaluation." Central to the drug-discovery approach at 3-Dimensional Pharmaceuticals (3DP), Exton, Pa., is the company's DirectedDiversity system, a strategy for computer-controlled robotic synthesis and analysis of chemical libraries to generate new drug leads. The company has large libraries of millions of druglike compounds that are registered in a database containing extensive chemical, structural, and synthetic descriptors. Using computer algorithms developed at the company, 3DP chemists formulate, test, and verify structure-activity models and select optimal sets of compounds from the database for targeted synthesis. The DirectedDiversity approach is used" to parallelize the process of structure-activity development, which remains a critical rate-limiting step in drug discovery," said 3DP Executive Director Richard M. Soll. Researchers at 3DP recently employed the strategy to identify potent inhibitors of thrombin (to prevent excessive blood clotting, a key factor in cardiovascular disease) and urokinase (to fight tumor metastasis). CombiChem has designed the Universal Informer Library of about 11,000 pure compounds so screening against any target is very likely to generate hits that can be used in the creation of more focused libraries, explained John Saunders, vice president for chemistry. The universal library is designed to be used for lead generation in cases where very little is known about the molecular target with which a potential drug will interact. Saunders says the company has already used its Universal Informer Library successfully in five drug-discovery projects. Researchers at Helios Pharmaceuticals, Louisville, use the company's Prolific Templates approach to construct libraries on highly functionalized molecular scaffolds, each representing a different chemical or structural characteristic. "Typically, a good chemist has been able to devise and optimize one to three combinatorial methods per year, consisting of five chemical steps on average, to produce libraries," said Adnan M. M. Mjalli, director of chemistry and automation at Helios. But with the Prolific Templates method, one Helios researcher recently "optimized the chemistry on solid support for the production of 15 individual libraries-each consisting of eight to 10 chemical steps and based on a different scaffold-in 15 weeks." Company scientists recently used the approach to create libraries of novel steroidomimetic compounds as potential sources of lead compounds for treatment of inflammation and diabetes. ArQule researchers work at an open-well chemistry synthesizer station. ArQule, Medford, Mass., specializes in solution-phase synthesis. "Most combinatorial libraries are still synthesized on solid supports, mostly due to the success in the peptide and nucleotide fields," said Alan P. Kaplan, director of lead optimization. "But we at ArQule feel that the future of combinatorial chemistry lies more in the realm of solution-phase parallel synthesis." ArQule's approach, said Kaplan, "is to blend medicinal and synthetic chemistry, parallel processing, and automation to be able to produce [libraries of] 250,000 unique and spatially addressed compounds as single-compound entities. While this number of compounds may be viewed as small compared to the number that can be made by split-and-mix solid-phase synthesis"-an efficient method for producing very large libraries-"we believe our approach overcomes many of the shortcomings of solid-phase synthesis." These include, he said, the limited range of reactions that can be run on solid supports, the necessity for encoding (identity-tagging) and deconvolution (identifying compounds by iterative library resynthesis), and the ambiguities involved in testing compounds as mixtures. ArQule researchers recently identified lead compounds from -ketoamide conjugate libraries and chalcone libraries that are active against protease targets, and they have generated some promising high-affinity ligands for key central nervous system receptors. Another distinctive strategy is that of Affymax Research Institute, which specializes in the construction of encoded solid-phase combinatorial libraries. According to Michael Needels, assistant director of combinatorial drug discovery:" Screening and hit identification of encoded libraries are faster and less labor-intensive than with unencoded libraries. Also, by encoding each member of a library, much less of each compound needs to be synthesized, so the size of the library can be much larger than is possible with unencoded libraries." Affymax is currently using encoded libraries for primary screening of a variety of targets. "In anticipation of preparing follow-up, or analog, libraries of initial hits," said Needels, "we feel it is important to design primary library chemistries so that every hit has a logical lead progression pathway. The initial euphoria over discovering a [potent] lead dissipates very quickly," he said, when it becomes apparent that it can't be developed into a drug. Combinatorial bio Companies also are using novel combinatorial biological and biosynthetic techniques to identify new drugs and biocatalysts. Todd Peterson, director of combinatorial biology at ChromaXome, San Diego, explained that in combinatorial biology or combinatorial biosynthesis, genomic DNA associated with metabolic pathways is manipulated and transferred from a variety of organisms into appropriate host microorganisms (such as bacteria) so that each individual organism receives a different set of genetic instructions. The host organisms proliferate and express the added biosynthetic genes, producing libraries of diverse natural products. These libraries are then screened to identify natural products associated with desirable biological activities. At Kosan Biosciences, polyketide synthase enzymes are used to produce libraries of novel polyketides, which the company refers to as "unnatural natural products." Polyketides exhibit an unusually broad range of biological activities (C&EN, March 9, page 29). In fact, the heavy-hitter pharmaceuticals belonging to this class fill a long list. "We are not trying to make hundreds of thousands of compounds," said Kosan's Santi. "Instead, we make very directed changes on molecules to improve their activity, like a microbial medicinal chemist might, or we make smaller libraries of focused molecules, in the thousands range." The goal is to identify polyketides and polyketide derivatives as lead compounds and new drugs. In addition to his role at Kosan, Santi is a professor of biochemistry and biophysics and of pharmaceutical chemistry at the University of California, San Francisco. He cofounded Kosan with chemical engineering professor Chaitan S. Khosla of Stanford University, who discovered the polyketide synthase technology on which the company is based. John R. Jacobsen, a postdoctoral fellow in Khosla's group, reported an application of polyketide bioengineering in which he was able to inactivate the biosynthesis of the erythromycin precursor 6-deoxyerythronolide B by site-directed mutagenesis of a key residue of bacterial polyketide synthase. He then added designed synthetic molecules to cultures of the mutated bacteria. The polyketide synthase module that normally recognizes 6-deoxyerythronolide B as a substrate for further biosynthetic processing also accepts these synthetic molecules as substrates. Hence, they were drawn into the biosynthetic pipeline, and a library of unnatural polyketides was produced. Further processing of some of these scaffolds by post-polyketide enzymes of the erythromycin pathway produced novel antibacterials with comparable in vitro potency to their natural counterparts, Jacobsen noted. Another technology used to create novel natural-product-like libraries is combinatorial biocatalysis, developed at EnzyMed. In this technique, enzymatic and microbial reactions are used to carry out the synthetic transformations needed to generate libraries. The resulting products are then assayed for useful biological and pharmacological activity. Chemical engineering professor Douglas S. Clark of the University of California, Berkeley, a cofounder of EnzyMed, noted that the company has used combinatorial biocatalysis to synthesize more than 200 derivatives of the anticancer agent paclitaxel-including a library of paclitaxel esters, some of which exhibit greatly improved water solubility compared to that of the natural product. Paclitaxel's sparing solubility in aqueous solution has caused problems in the administration of the drug. "EnzyMed is the first company to integrate enzymatic, microbial, and chemical reactions for combinatorial synthesis," said Clark. "This approach can be applied to a wide range of substrates, from simple precursors to very complex molecules. Complementary research on combinatorial biocatalysis, including implementation of solid-phase methods, is under way in my own lab and that of my collaborator, Jonathan S. Dordick," professor of chemical and biochemical engineering at the University of Iowa, Iowa City, and a cofounder of EnzyMed. Researchers at Transcell Technologies examine a carbohydrate-based combinatorial library arrayed in 96-well microtiter plates. Transcell Technologies, Cranbury, N.J., is focusing specifically on the generation and assaying of carbohydrate libraries. The company uses a glycosylation technique developed by its founding scientist, chemistry professor Daniel E. Kahne of Princeton University, to form glycosidic linkages to various types of molecules. "Using unique sugar building blocks, we can explode the diversity around the sugar nucleus," said Michael J. Sofia, Transcell vice president of research and director of chemistry. "We build directed libraries around some interesting natural products that have sugars associated with them, and to construct primary screening libraries we use sugars as templates or scaffolds for displaying functional groups in a three-dimensional orientation," said Sofia. He pointed out that the company's design strategy for making screening libraries takes advantage of early concepts developed by chemistry professor Ralph F. Hirschmann of the University of Pennsylvania for the use of carbohydrates as peptide mimetics. Miniaturization As applications of combinatorial chemistry proliferate, the ability to keep equipment size and reagent quantities to a minimum is an issue of growing importance. Stephen Felder, a partner at NeoTech Instrumentation, Tucson, described the advantages of a miniaturized combinatorial synthesis and screening system that the company will offer for the preparation and evaluation of libraries of up to 30,000 discrete compounds. The system includes a unit that dispenses prepared compounds to assay plates in an automated manner and a miniaturized screening unit for carrying out biochemical and cell-based assays in cells with volumes as small as 2 µL. Also new under the combinatorial sun is a microfluidic chip-based system that is being developed by Orchid Biocomputer. According to Orchid's DeWitt, the company's five-layer, 2-inch-square chips can be used to carry out chemical reactions at submicroliter volumes in 144-well arrays. "Why miniaturize?" DeWitt asked. Small-scale equipment makes it possible, she said, to use less sample, take advantage of lower reagent costs, improve analytical sensitivities, and integrate chemical synthesis seamlessly with other drug-discovery activities, including high-throughput screening and genomics. Reagent savings are particularly important. DeWitt pointed out that the reagents required to create a 10,000-member library of Dilantin analogs would cost more than $200,000 with a popular commercial solid-phase organic synthesis system, but less than $100 with an Orchid microchip. Pumping fluids through the chip is either electrohydrodynamic (for organic solvents) or electroosmotic (for aqueous solutions). Reactions currently are performed on commercially available solid-phase synthesis beads (one to five per well), but solution-phase synthesis also should be eventually possible. Orchid has a collaboration with SmithKline Beecham to develop the capability to perform 10,000 simultaneous reactions on microfluidic chips for drug-discovery applications. Another lab-on-a-chip device, in this case specifically for use in high-throughput screening, is being developed by Caliper Technologies, Palo Alto, Calif. The movement of solvents and solutions through Caliper's two-layer chip is achieved by electroosmotic pumping. According to Caliper Vice President J. Wallace Parce: "Assays performed in microfluidic devices can be very reproducible. When such devices are coupled with pipetters, they can be used to perform high-quality, high-throughput assays on large libraries of compounds." Another miniaturized high-throughput screening system has been developed by Axiom Biotechnologies, San Diego. The company's HT-PS system uses computer-controlled valve mechanisms and fluorescence detectors to monitor the effects of library compounds on cell receptors and ion channels via measurements of intracellular ion concentration changes. The system currently carries out assays in microliter volumes, "and this should be scalable down to the nanoliter assay level over time," said Mark E. Goldman, Axiom senior director for drug discovery. In addition to pharmacological evaluation, the system is designed to perform functional, potency, and selectivity profiling of hit compounds. Combinatorial applications Of course, the main purpose of combinatorial chemistry is to discover things, and speakers at the meetings indicated that new drugs and materials are indeed being discovered combinatorially. Acadia Pharmaceuticals, San Diego, is one of several companies that are focusing their combinatorial armamentarium on genomic targets. Acadia's technology is based in part on work carried out in the 1980s at the National Institutes of Health by Mark R. Brann, company founder, president, and chief scientific officer. At NIH, Brann was among the first to discover muscarinic acetylcholine receptors and other drug targets by gene cloning. Brann says he has learned since then that it's actually a lot easier to discover potential drug target genes than it is to test the effects of drugs on those targets. Nevertheless, Acadia has developed high-throughput screening procedures that seem to be up to the job. Brann reported some promising data the company has obtained on G-protein receptors, the JAK/STAT signal transduction pathway, and tyrosine kinase linked receptors. Combinatorial efforts to hit G-protein receptor targets also are being pursued at Synaptic Pharmaceutical, Paramus, N.J. Stewart A. Noble, vice president of chemistry, described small-molecule libraries constructed as sources of new or improved drugs for G-protein receptor-related disorders and also as probes of receptor function. G-protein receptor targets for which libraries are being designed and constructed at the company include NPY receptors (which play a role in obesity, anxiety, diabetes, and analgesia) and galanin receptors (which are involved in obesity, cognition, pain, and diabetes). James R. Hauske, senior vice president for discovery, and his coworkers at Sepracor, Marlborough, Mass., are using combinatorial chemistry to identify novel agents effective against antibiotic-resistant strains of Staphylococcus aureus bacteria (which cause pneumonia) and Enterococcus faecium bacteria (which cause blood poisoning). The researchers have developed a series of novel organic reactions on solid supports to generate libraries of substituted heterocycles. Two lead compounds identified from these libraries have shown antibiotic activity approaching that of the commercial antibiotic vancomycin in studies on infected mice. Hauske and coworkers also have developed intermolecular ring-opening cross-metathesis reactions for the regioselective and stereoselective synthesis of highly functionalized carbocycle and heterocycle libraries. Antibacterial agents also are being sought by researchers at Versicor, Fremont, Calif. There, molecular targets are grouped in families (based on commonalities in mechanism of action) to facilitate the selection of pharmacophores capable of inhibiting those targets. The researchers develop solid-phase methodologies to prepare libraries of different druglike scaffolds bearing the pharmacophores and then screen them for activity. Dinesh V. Patel, senior director for chemistry, and his coworkers at Versicor recently prepared libraries in which a hydroxamate pharmacophore was linked to various cyclic and acyclic scaffolds. From those libraries, they were able to identify some potential inhibitors of metalloproteases, and this led to the discovery of some promising antibacterial lead compounds. Patel also described the company's lead optimization efforts on oxazolidinones, a class of antibacterial agents currently in clinical trials. In addition, Jeff W. Jacobs, associate director of chemistry, explained the company's approach to high-throughput parallel synthesis, in which dedicated instrumentation is used to automate resin dispensing, reagent dissolution, compound characterization, and other steps that tend to slow up the synthetic process. Metalloprotease inhibitors are likewise a key focus of interest at Affymax Research Institute. "Companies have had matrix metalloprotease inhibitors in clinical trials, but some of the drugs had to be dropped because of side effects, and so far no compound has really made it to the market," explained Affymax senior scientist Anna Katrin Szardenings. Many of the drugs tested so far were peptidelike, she said. Such compounds are easy to obtain and they make potent inhibitors, but they have often exhibited some unfavorable properties, such as poor bioavailability. Based on a pharmacophore Szardenings developed, libraries of potential ligands based on 2,5-diketopiperazine heterocyclic scaffolds were synthesized. Several compounds from those libraries have shown good potency profiles as matrix metalloprotease inhibitors, she said, and" we expect the properties of these compounds to be much more favorable than those of peptides." A popular strategy is to combine molecular modeling and rational-design strategies with combinatorial chemistry methods in pharmaceutical research. For example, Labaudiniere of Rhône-Poulenc Rorer said methodologies have been developed at his company "to design diverse libraries and biased libraries, essentially using methods involving systematic 3-D pharmacophore analysis." Researchers at Molecumetics, Bellevue, Wash., also combine rational design and combinatorial chemistry in their drug-discovery program. According to Michael Kahn, chief scientific officer at Molecumetics and professor of pathobiology at the University of Washington, Seattle: "We look at structural biology and at the particular secondary-structure elements that an enzyme or receptor really uses for recognition. That's basically conserved within superfamilies. For example, all proteases and kinases and most SH2 domains look at extended ß-strand structures when they interact with their substrates or ligands. For a particular problem, we make libraries that have that superfamily motif embedded in them." Molecumetics project leader John Tulinsky noted that some of the company's compounds recently moved into preclinical testing. Thrombin inhibitors (for use as cardiovascular therapeutics) are most advanced in the development cycle, but inhibitors of tryptase, urokinase, and a host of other proteases are in the wings. Symyx scientist loads a thin-film library into a scanning mass spectrometer to screen for catalytic activity. At Symyx Technologies, Santa Clara, Calif., scientists use combinatorial chemistry not to develop new drugs, but to create inorganic materials with useful electronic, magnetic, and optical properties (C&EN, Dec. 8, 1997, page 24). According to W. Henry Weinberg, chief technical officer at Symyx and professor of chemical engineering, materials engineering, and chemistry at the University of California, Santa Barbara, "The application of combinatorial synthesis and screening for the discovery and understanding of classes of materials that are not necessarily pharmaceuticals is potentially of greater impact than the already maturing drug-discovery applications." At Symyx, thin-film techniques such as physical vapor deposition are used to produce 3-inch-diameter silicon wafers having up to 25,000 different material compositions on their surfaces. By testing such materials, company researchers recently discovered a novel blue-white phosphor that will potentially find applications in flat-panel displays, fluorescent lights, and computer screens [Science,279, 837 (1998)]. Currently, the company also is working on a project to find new olefin polymerization catalysts. "Last year's debate was, Will the technology work?" said Weinberg. "But this year's debate might be, What are the best applications of the methodology? In other words, I think people are now becoming aware of the fact that this is going to more or less revolutionize the way materials science is done. It's a question of how big the revolution will be, as opposed to whether there will be one. Applications include electronic materials generally, certainly catalysis, and I would also add macromolecular, polymeric, and organic materials." Charybdis Technologies, Carlsbad, Calif., currently is planning to launch a chemical research division called Scylla Chemical that will be focused on another decidedly nonbiological target-organometallic catalysts. "In the Aldrich chemical catalog there are about a dozen different catalysts for performing hydrogenations," said Thomas J. Baiga, president, chief executive officer, and cofounder of Charybdis. "Each has its own specific area where it works really well, and these catalysts evolved over tens of years to get this much selectivity. Now we're talking about being able to do this with combinatorial chemistry in a six- to 12-month period. It's a tremendous advantage." Also in search of new organometallic catalysts is chemistry professor Kevin Burgess at Texas A&M University, College Station. Burgess' group favors a method in which small libraries of potential catalysts are made by parallel synthesis and then screened sequentially. New supports New solid supports and solid-phase techniques also continue to emerge for use in combinatorial synthesis. For instance, Solid Phase Sciences Corp., Watertown, Mass., has developed magnetic supports as a potential replacement for the filtration units commonly used in automated solid-phase synthesis and separation instruments. The company's layered polystyrene-coated paramagnetic beads have high loading capacities, meaning they'll hold a good amount of synthesized compound. Although magnetic supports are more expensive than nonmagnetic ones, magnetic beads will be cost-effective when used to automate large-scale syntheses, where filtration can be troublesome because of clogging and other problems, according to Irving Sucholeiki, the firm's founder and president. At the meetings, Kimberly G. Estep, senior research investigator at Pfizer Central Research, Groton, Conn., made the first public disclosure of a promising new "clean-break" or "traceless" linker for solid-phase synthesis. According to Estep, "One of the disadvantages of solid-phase chemistry is that the final products often retain the chemical functionality that was used to link the product to the resin"-a potentially undesirable constraint on the structure of the products. Her new linker can be prepared in two simple steps from commercially available starting materials. "Products are obtained cleanly and in high yield using mild cleavage conditions and bear no chemical evidence of attachment to a support," Estep noted. The linker has been widely adopted by chemists at Pfizer for solid-phase syntheses of secondary amides, ureas, sulfonamides, guanidines, anilines, and other compounds, she said. Senior research scientist Rongshi Li of Irori also reported on the development of improved supports. Irori's proprietary combinatorial chemistry system includes two types of microreactors: MicroKans, small solvent-permeable microreactors loaded with solid-phase beads on which synthesis takes place, and MicroTubes, reactors whose surfaces are used as sites for solid-phase synthesis. An encoding device is placed in each microreactor to identify the synthetic product produced there. Li is currently developing potential acid-labile linker chemistries to create new types of MicroTubes to supplement the four kinds currently available commercially. Analysis A crucial aspect of combinatorial chemistry is being able to identify the compounds produced in a combinatorial synthesis. Hence, considerable efforts are being devoted to the chemical analysis of combinatorial libraries, as was clear at last month's Pittsburgh Conference in New Orleans (C&EN, March 16, page 42). The reason analysis is so important, said Czarnik, is that "not all reactions work. What you really would like to know is, not just what reactions were run, but are the compounds in this library what I think they are and how pure are they? You don't know what you have until you characterize your library properly." However, he said, "there's a lack of consensus on how best to analyze a compound library. The methods people are using today to analyze the quality of compounds in a library aren't really very good. "I think one of the exciting developments in this area is work on high-throughput NMR," Czarnik said. "It is now possible to autosample plates of compounds, pass them through an NMR probe, and get an NMR spectrum on the compounds in three minutes. A lot of scientists wrote NMR off because the way it's done classically is such a slow procedure. But people who are working on the high-throughput NMR method are showing it's possible to use it for combinatorial analysis." Bing Yan, senior associate fellow at Novartis Pharmaceuticals, East Hanover, N.J., reported on infrared spectroscopy techniques for use on individual solid-support beads and methods for on-bead quantitation of reaction products. "The ability to do multiple steps and drive each reaction to completion are major advantages of solid-phase synthesis," said Yan. "But it's very important to confirm that every step is complete. Up to now it's been very hard to do this because you had to cut compound off the bead to analyze it. Now we've developed methods to identify and quantitate reaction products right on the bead. We don't have to interrupt the reaction. This makes reaction optimization of solid-phase synthesis much easier, especially for multistep synthesis." Yan and coworkers also have demonstrated the ability to obtain data routinely on the kinetics of solid-phase organic reactions by single-bead IR analyses. Analytical instruments also are being customized for combinatorial applications. For example, product manager Janice A. Ramieri of the Biotage Division of Dyax, Charlottesville, Va., described the Parallex HPLC, a semipreparative multichannel high-performance liquid chromatography system for library purification. "The idea that you can do a five-stage synthesis and end up with pure compounds just hasn't worked," explained Ramieri's colleague, Biotage Vice President Patrick Coffey. "You get by-products and incomplete reactions along the way." Once a library has been synthesized, he said, "you can take it into an assay dirty, but that causes false positives, false negatives, and lots of other problems. The Parallex HPLC can be used in an automated fashion to take synthesized compounds and purify them by UV monitoring of separate flow streams. It's designed to handle up to about 250 compounds in an eight-hour shift." Automation Considerable effort also is being devoted to automating combinatorial synthesis." The ability to synthesize small-molecule libraries [having] tens of thousands of members is becoming a standard job expectation in laboratories engaged in drug discovery," said Czarnik, "and the sheer numbers involved demand that reaction vessel handling become automated." According to Vaisberg: "Over a dozen manufacturers worldwide now offer off-the-shelf semiautomated and fully automated equipment for both solution- and solid-phase parallel synthesis. Taking into account that the price range for that kind of equipment starts at a few thousand dollars and that in many cases even simple semiautomated equipment can ensure sufficient productivity, parallel synthesis methods are now broadly accessible to any industrial or even academic research group." One of the major changes in the combinatorial market this year is that major analytical instrument vendors are beginning to move in. "Combinatorial chemistry is now well in the sights of the big instrument companies," said Charybdis' Baiga." This is the year that we expect to see both Hewlett-Packard [Palo Alto, Calif.] and Perkin-Elmer Applied Biosystems [Foster City, Calif.] start taking an active role in this. And it's going to be very interesting to see how small-niche automation companies like ours are going to fare against these larger companies." Indeed, said Houghten, "there's tremendous activity and innovation in the synthesizer and robotics area, and what's going to happen is the usual thing-the early people aren't necessarily going to be the successful ones. There will be a shakeout here, I think, fairly soon." At the Cambridge Healthtech meetings, Hewlett-Packard presented the HP 7686 Solution-Phase Synthesizer, an instrument that can automate a range of chemistries, including Grignard, Wittig, Diels-Alder, and Heck reactions, Suzuki coupling, nucleophilic aromatic substitution, Michael addition, and heterocyclic synthesis. Paul D. Hoeprich Jr., marketing program manager, explained that the system automates reaction setup and incubation, compound workup procedures, and analytical sample preparation methods. Solid-phase "scavenger" cartridges are included to facilitate the selective removal of unreacted starting material from solution. "HP's approach," said Hoeprich, "is to provide automation that is compact, affordable, and easy to use so individual chemists can have their own personal productivity tool." He said future HP synthesizers will automate a broader range of chemistries and will provide tighter integration with analytical equipment and more efficient information management. In addition, Perkin-Elmer Applied Biosystems is developing an automated solid-phase synthesizer. "We're hoping for a fall '98 release," said group leader B. John Bergot. "We're in partnership with Tecan [Hombrechtikon, Switzerland] to develop a unit that's loosely based on Tecan's CombiTec automated parallel synthesizer, but with significant enhancements and upgrades. It's got a 48-position reactor block, for which patents were issued in January. With another party, which I'm not at liberty to identify at the moment, we are also developing the ability to link up our device with a sophisticated data management software system." According to Bergot: "A fully outfitted robot is a fairly expensive piece of capital equipment. What we want to try to extend to users is a multiuser concept in which several researchers will have use of the deck itself, and then the reaction blocks are quite portable, so they can be demounted and removed to other locations for incubation, refluxing, and that sort of thing. You can imagine several blocks in the hands of different researchers who will have their own particular chemistries to pursue." The system, added Bergot, "is for small-molecule synthesis of defined libraries. Several types of organic reactions have already been demonstrated on it, including the Suzuki, Heck, and Mitsunobu cross-coupling reactions and butyllithium metallation. It will also support optimization studies and small-scale process development." Linda Hamaker, senior applications scientist at Advanced ChemTech, Louisville, said she is seeing a trend "toward lower cost semiautomated, as opposed to fully automated, products. Our company as well as others have fully automated systems, but they're very expensive for some organizations, such as universities and start-up companies. Especially if they've already got manual synthesis down to a tee, they just want to make it a little faster. Probably last year you wouldn't have seen any ads for anything small and semiautomated, but this year they're all over the place." With Advanced ChemTech's ReacTech semiautomated organic synthesizer, said Hamaker, "you use a syringe to add your reagents, but all the washing is automated. You can walk away and add a pause step when you need to add your reagents, then let it go again. The instrument can do both solid- and liquid-phase synthesis." Although it is relatively inexpensive, the system includes a 40-well reactor, a reaction heating and cooling system, variable-speed mixing, inert nitrogen or argon reaction atmospheres, and other features. "Whether you adopt a solid- or solution-phase approach to your synthetic target, your first and foremost aim is optimization of the reaction prior to library synthesis," said Alasdair A. MacDonald, worldwide applications scientist at Argonaut Technologies, San Carlos, Calif. "Using Argonaut's Nautilus 2400 automated synthesizer, you can quickly optimize reaction conditions such as time, temperature, reagents, solvents, and concentration," he said. The Nautilus provides an inert environment for solid- and solution-phase synthesis and makes it possible to control temperature, time, and reaction conditions. Argonaut also offers the Quest 210, a relatively inexpensive instrument for parallel synthesis of 20 compounds per run, also by solid- or solution-phase methods. The company is developing an instrument for high-throughput synthesis of 192 compounds per run. Charybdis Technologies' synthesizer, the Calypso system, also is designed to automate solution- and solid-phase parallel synthesis. The instrument has a modular reaction block that can be configured in a 24-, 48-, or 96-well format. Temperature is controlled by circulating fluids, and the reaction atmosphere is controlled by a gas manifold system. Charybdis also manufactures a liquid-handling robotic workstation for combinatorial synthesis, the Iliad PS², and will soon release a wash/cleavage workstation (Phorcys PS²) and a parallel purification system (Syrinx PS²). FlexChem, a modular system for high-throughput chemistry introduced by Robbins Scientific, Sunnyvale, Calif., is designed to prepare libraries of discrete compounds. Sales and marketing director James E. Stanchfield explained that the system consists of a 96-well synthesis and filtration block that can be sealed independently for carrying out solid-phase chemistry. The 96 reaction chambers are arrayed in a standard microtiter plate format for compatibility with robotic liquid-handling systems. A production-level automated synthesizer for solid-phase parallel synthesis is also being introduced by Zeneca Pharmaceuticals, Wilmington, Del., which developed it in collaboration with Zymark, Hopkinton, Mass. The synthesizer features an array of 96 reaction vessels and includes capabilities for parallel addition and parallel clearance of reagents and solutions to and from reaction vessels. "Some automated platforms will not allow you to perform certain synthetic operations, so you may not be able to prepare specific targeted compounds," said James B. Campbell, Zeneca's assistant director of chemical technology. "A particular goal we had was to minimize that compromise." Irori is introducing a sorting system, the AutoSort-10K, that automates split-and-mix solid-phase organic synthesis when used in conjunction with the company's microreactors system. "Most pharmaceutical companies want to make every compound in a library in significant amount," said Czarnik. "The way to accomplish that is not to do random splitting, but to provide every compound with a schedule, so every one of 1,000 microreactors sees a different series of reactions. When that occurs you get a different product for every one." The slow step in that process," he said," is taking a microreactor, reading its code from the radio-frequency chip that's inside, and putting it into a flask. You can sort about 2,000 in a day if you're really rugged, so sorting 10,000 is going to take a week of doing nothing but sorting. Nobody wants to do this." That's why the company developed the AutoSort-10K, which can read 10,000 microreactors and move them to individual reaction vessels in 10 hours. Senior fellow Árpád Furka of Helios Pharmaceuticals, who invented split-and-pool synthesis in 1982 (when he was an organic chemistry professor at Eötvös University, Budapest, Hungary), commented that Irori's system "will outperform parallel synthesis by several orders of magnitude while keeping all of its advantages." In addition to the instruments discussed in presentations at the Cambridge Healthtech Institute meetings, a number of other automated synthesizers are commercially available, including the RAM system (from Bohdan Automation, Mundelein, Ill.), the Myriad Personal Synthesizer (from the Technology Partnership, Melbourn, Royston, England), and the CombiTec system (from Tecan). According to Czarnik: "The motivating theme for combinatorial chemistry might be to be able to make a million variants of any structural template and screen them. Cheaper and faster are better. When we can accomplish this within a week, then perhaps the intellectual challenges will be gone. However, I can assure you that we are far, far from that vision today." Czarnik said he agrees with Joseph C. Hogan Jr., chairman of the board and chief scientific officer of ArQule, that essentially all of organic chemistry deserves to be reworked to be useful in a high-throughput setting. "Think about it," said Czarnik. "In a combinatorial scheme, every reaction must work in high yield and with a variety of reagents. That's an onus only the peptide and oligonucleotide chemists have had to face previously. The oligosaccharide crowd is facing it now. Literally, all of organic chemistry can be reworked with this goal in mind. There's a lot of work to be done!"

Industry is embracing the technology 'totally,' as researchers continue to advance the art of rapid synthesis and screening