STU BORMAN, C&EN WASHINGTON

COMBI CATALYSTS In a combinatorial study to optimize a homogeneous catalysis reaction, researchers plotted results from screening various conditions for a palladium-catalyzed Heck reaction. Horizontal axes show different solvents and bases used in the reaction, vertical displacement of spheres represents percent conversion of substrate to product, and sphere size indicates level of undesired palladium black formation indicative of catalyst deactivation. DMA = dimethylacetamide, DMSO = dimethyl sulfoxide, TEA = triethylamine. COURTESY OF CYNORA GMBH

Combinatorial chemistry--the synthesis of chemical compounds as ensembles (libraries) and the screening of those libraries for compounds with desirable properties--continues to evolve as a potentially speedy route to new drugs, catalysts, and other useful compounds and materials. Yet the field is facing increasing criticism for not having delivered results fast enough to justify the huge investments made in it by industry, academia, and government.

At a number of conferences held this year, researchers debated that issue and discussed advances in several areas, including natural-product-like libraries, dynamic combinatorial chemistry, combinatorial optimization of catalysts, and multicomponent reactions.

"The subtleties of nature do not allow us to easily design compounds with perfect properties," says Morten Meldal, leader of the Center for Solid Phase Organic Combinatorial Chemistry at Carlsberg Laboratory in Copenhagen. He chaired Eurocombi 2--the 2nd European Symposium on Combinatorial Sciences in Biology, Chemistry, Catalysts & Materials--held this summer in Copenhagen.

The field "is still in its infancy, together with genomics and proteomics," Meldal tells C&EN. It has already had a dramatic influence on the chemical discovery process, "but it still has a long way to go before its full potential can be realized."

"Combinatorial chemistry is useful in drug discovery, it can be an intellectually stimulating activity, and it does create value to society," said Anthony W. Czarnik, editor of the Journal of Combinatorial Chemistry, at an American Chemical Society ProSpectives conference on combinatorial chemistry last month in Leesburg, Va. "I don't think there's anyone today who would stand up and put their name and reputation behind the statement that combinatorial chemistry is useless."

By applying the tools of combinatorial synthesis to make and screen large numbers of molecules, "we're coming up with new leads that could not have been found before simply by screening natural products," Czarnik continued. He cochaired the ProSpectives conference with associate professor of chemistry Michael G. Organ, who is director of the Combinatorial Chemistry Facility at York University, Toronto, and reader A. Ganesan of the Combinatorial Centre of Excellence at the University of Southampton, England.

"Has combinatorial chemistry delivered on providing compounds with important properties?" asked University of California, Berkeley, chemistry professor Jonathan A. Ellman at the same meeting. "I would argue that it has delivered in providing useful and important compounds. The concept of making compounds in parallel and testing them in parallel to more rapidly obtain new properties--it's very clear that this has played an important role and will continue to play an important role in a vast range of disciplines."

Synthesizing or purchasing libraries of purified, discrete compounds has become the norm at many drug companies, Ellman said. "There are certainly applications where mixtures can be useful, but it's clear that discretes are primarily what are produced and that purification and characterization are highly valued."

FOR THE MOST PART, libraries are being made either by parallel synthesis or by mixture synthesis with "directed sorting," an approach used to keep track of individual compounds. Split-and-mix synthesis, a technique for synthesizing compounds in large numbers and with great diversity, "has dropped to a more minor contributor, relative to these other strategies," Ellman notes.

DIVERSITY In hapalosin analogs synthesized by Maier and coworkers (bottom), a variety of amino acid side chains are incorporated at R1, and aldol condensation is used to add a range of functional groups at R2 and R3.

Meldal agrees that an initial emphasis on creating mixtures of very large numbers of compounds "has largely given way in industry to a more measured approach based on arrays of fewer, well-characterized compounds." And he notes a particularly strong move toward the synthesis of complex natural-product-like compounds--molecules that bear a close structural resemblance to approved natural-product-based drugs.

Jeremy K. M. Sanders, head of the department of chemistry at Cambridge University, says that "very large libraries of a million compounds or more are going out of fashion, but smaller, more focused libraries are still much in evidence. The main dividend of combichem in pharma is that it provided the stimulus for robot-controlled and immobilization strategies that allow high-throughput and multiple parallel approaches to drug discovery. Combinatorial approaches are also now being adopted for solid-state and materials applications and in the search for new catalysts."

Professor of medicinal chemistry Rob M. J. Liskamp of Utrecht University, in the Netherlands, adds that "combinatorial chemistry is now firmly established as an important tool in drug discovery--not so much for synthesizing and screening huge libraries, but for all the combinatorial tools that have been developed, such as solid-phase synthesis, resins, reagents, linkers, and clever screening methods. The field does not have to demonstrate its value any more, and expectations for it are now at a realistic level."

When molecular modeling was initially developed in the 1980s, Liskamp says, "it was presented as the ultimate solution for designing every desired ligand. After a while, people saw the pros and cons of the methodology more clearly, and it assumed a more realistic, albeit still essential, status. I think the same has happened now with combichem."

Sanders and Liskamp were plenary and invited speakers, respectively, at Eurocombi 2. The meeting was arranged by the European Society of Combinatorial Sciences (ESCS), an organization that some hope will change its name and expand into the first international nonprofit organization dedicated to combinatorial chemistry.

Earlier this month, the National Institute of General Medical Sciences (NIGMS), a unit of the National Institutes of Health, renewed an earlier vote of confidence in combinatorial technology by awarding a total of $5.6 million in first-year funding to two new Centers of Excellence in Chemical Methodologies & Library Development (CMLDs).

The first two CMLDs--at Boston University and the University of Pittsburgh--were funded in September 2002 (C&EN, Nov. 11, 2002, page 43). Two more are now being set up: One, at the University of Kansas, Lawrence, focuses on the design of molecular scaffolds (combinatorial core structures) and is led by medicinal chemistry professor Jeffrey Aubé. The other, at Harvard University, specializes in diversity-oriented synthesis and is headed by chemistry and chemical biology professor and Howard Hughes Medical Institute investigator Stuart L. Schreiber. NIGMS estimates that the CMLD awards will total nearly $40 million over the five-year initial grant lifetimes of all four centers.

"Together with an array of recent advances in biomedicine, the broad availability of diverse chemical libraries could help launch a new and exciting era of preclinical disease detection and personalized and targeted medicines," said NIH Director Elias A. Zerhouni in announcing the two new CMLDs.

The topic of "molecular libraries and imaging" is also one of five "new pathways to discovery" highlighted in the NIH road map announced last month (http://nihroad map.nih.gov). As part of this program, NIH will assemble a huge combinatorial library as a source of new drug candidates.

According to NIH, "National centers will initially establish a collection of 500,000 chemically diverse small molecules of both known and unknown activities. Over time, this collection will be expanded and modified to provide a working set of compounds that will target larger domains of 'biological space,' the total set of biomolecular surface domains that are capable of interacting with a small molecule."

The library will "offer public-sector biomedical researchers access to small organic molecules [that] can be used as chemical probes to study cellular pathways in greater depth." The initiative will also "speed the development of new drugs and agents to definitively detect and treat common and rare diseases by providing early-stage compounds that encompass a broad range of novel targets and activities."

NATURAL-PRODUCT-LIKE LIBRARIES. Pharmaceutical companies have now been using combinatorial chemistry for drug discovery for about a decade. Some of the earlier libraries they synthesized have been discredited for being poorly designed, impractically large, and structurally simplistic. That's why drug researchers are increasingly embracing natural-product-like libraries--moderate-size collections of complex compounds that are highly likely to exhibit interesting and useful types of biological activity.

	THREE-ARMED Triazacyclophane-based tripodal "receptor" with peptidic arms. Liskamp and coworkers are creating libraries of such tripodal compounds as a first step toward small-molecule mimics of antibodies. "We will ultimately decorate the scaffolds with cyclic peptides to mimic the loop structure of antibody complementarity-determining regions," Liskamp says. Note: R1 to R9 are amino acid side chains.

At the ACS ProSpectives conference, researcher Michael A. Foley of Infinity Pharmaceuticals, Cambridge, Mass., pointed to a number of problems with earlier combinatorial libraries. They were often based on a single skeleton, the individual library members were structurally similar, only a limited number of skeletons were accessible, and compounds tended to be predominantly achiral or racemic. Such compounds couldn't reach many biological target types, he said.

Foley and colleagues use diversity-oriented synthesis to design complex libraries that will potentially prove to have fewer drawbacks. They use biologically relevant building blocks and branching networks of reactions to produce libraries of natural-product-like compounds that have properties consistent with those of drug leads. Infinity is currently screening such libraries for compounds with antibacterial, antifungal, and anticancer activity.

A potential step toward natural-product-like libraries of unprecedented diversity was made last week, when Schreiber and coworkers reported a strategy for making small-molecule libraries with all possible combinations of a set of both core skeletal structures and peripheral groups [Science, 302, 613 (2003)]. In most library syntheses, diverse peripheral groups are attached to a single core. The new approach achieves core and peripheral diversity simultaneously by using special core structures ("latent intermediates") that react with peripheral groups ("skeletal information elements") to generate diverse skeletons.

Chemistry professor Martin E. Maier and coworkers at the University of Tübingen, Germany, have been using parallel synthesis to create natural-product-like libraries of cyclic depsipeptide analogs. Depsipeptides are peptidelike compounds with both ester and amide linkages instead of just amide bonds. Maier's group has synthesized libraries of hapalosins, depsipeptides with known activity in reversing multidrug antibiotic resistance. Some hapalosin analogs they made have turned out to be biologically active. The researchers are currently developing methods to isomerize aldols and other small-molecule building blocks, combine them, and then cyclize them to form depsipeptides--a method that mimics the way terpenes are synthesized biologically.

Cyclic depsipeptides have also been synthesized by other groups, including those of chemistry professors Richard B. Silverman of Northwestern University, Dale L. Boger of Scripps Research Institute, and Takashi Takahashi of Tokyo Institute of Technology. Last year, Boger and coworkers synthesized a library of cyclic depsipeptides and developed a "chemical mutagenesis" approach to defining their structure-function properties. And following up on their total syntheses of the natural depsipeptide sansalvamide A and the natural cyclic peptide scytalidamide A, Silverman and coworkers recently made a library of sansalvamide A analogs and found two of them to be more potent than the natural product.

Chemistry professor Sidney M. Hecht and coworkers at the University of Virginia, Charlottesville, have been synthesizing bleomycins and bleomycin analogs, both as mechanistic probes and potentially bioactive agents. Bleomycins are glycopeptide antitumor antibiotics that are believed to act by a sequence-selective nucleic acid cleavage mechanism and are used to treat squamous cell carcinomas, malignant lymphomas, and other cancers. Hecht's group recently synthesized, purified, and characterized a library of 108 deglycobleomycin analogs [J. Am. Chem. Soc., 125, 8218 (2003)]. The study was the first in which analogs of bleomycin or deglycobleomycin were found to have greater DNA cleavage activity than the parent bleomycins from which they were derived.

Liskamp and coworkers are designing tripodal (triple-binding) agents "to obtain molecular constructs with levels of affinity and selectivity similar to those of antibodies," Liskamp explains. The antigen-binding site of an antibody is composed of six hypervariable loops, or complementarity-determining regions (CDRs). "We use relatively small synthesized molecules as scaffolds that each present three different peptide chains, or loops," he says. "Two such sets of loops, if covalently combined, would lead to a molecular construct containing six loops, like an antibody's CDRs. The idea is to have six different ligands interacting with different sites on the same antigenic protein."

Last year, Liskamp and coworkers took a key step toward this goal by decorating a triazacyclophane scaffold with three different peptides [J. Comb. Chem., 4, 275 (2002)]. More recently, they showed that cyclotriveratrylene can be used as a scaffold for the solid-phase synthesis of a library of tripodal "receptors" with peptidic arms [J. Comb. Chem., published online Aug. 26, http://dx.doi.org/10.1021/cc034003u].

DYNAMIC COMBICHEM. Dynamic combinatorial chemistry (DCC) is a relatively new way of synthesizing and identifying small molecules that bind with high affinity to macromolecular receptors--or conversely, synthetic receptors that bind tightly to small molecules. DCC uses equilibrium forces to amplify compounds that bind well to targets and discriminate against compounds with lower affinity.

Full-sized Image

Early demonstrations of DCC were made by several research groups, including those of Sanders at Cambridge University; Director of Chemistry Alexey V. Eliseev at Therascope AG, Heidelberg, Germany; chemistry professor Jean-Marie Lehn of Louis Pasteur University, Strasbourg, France; retired Columbia University chemistry professor W. Clark Still; and associate professor of dermatology Benjamin L. Miller at the University of Rochester Medical Center. But more recently, the number of practitioners has been broadening considerably and the number of applications has been growing. DCC is a field "of which you will be hearing much more in the next few months," Sanders says.

Senior research fellow Sijbren Otto of the department of chemistry at the University of Cambridge told ACS ProSpectives attendees that he and his coworkers have been using reversible disulfide chemistry to prepare DCC libraries of varied macrocyclic compounds. They then screen the macrocyclic "receptors" against small molecules to find compounds that bind tightly. Otto noted that DCC can also be used to find macrocyclic receptors with catalytic activity--by screening macrocyclic libraries against guests that resemble transition states of reactions.

Miller and coworkers have recently been trying to extend DCC to bioincompatible reactions. They're using an idea borrowed from dialysis--membrane-based aqueous- and organic-phase compartmentalization--to set up DCC systems with biphasic equilibrium. "Although we're still at the early stages of doing this," Miller tells C&EN, "results thus far suggest that we will be able to significantly extend the range of chemistry accessible to the DCC-based identification of biomacromolecule-binding small molecules."

Associate professor of biochemistry Romas J. Kazlauskas and coworkers at the University of Minnesota, St. Paul, and associate professor of chemistry James L. Gleason and coworkers at McGill University, Montreal, are collaborating to develop "pseudodynamic" combinatorial techniques. These address a key drawback of DCC technology: fundamental limitations on the degree of selectivity or amplification that can be achieved, owing to the thermodynamics of the equilibrium processes used in DCC.

In pseudodynamic combinatorial chemistry, an irreversible destruction reaction is used to destroy poor binders kinetically by removing them from solution. When compounds bind strongly to a target, they're sequestered from the destruction reaction. They thus get "selected" and concentrated as poorer binders are depleted.

Last year, Kazlauskas, Gleason, and coworkers showed that an irreversible protease-catalyzed hydrolysis reaction could be used to select and concentrate aryl sulfonamide dipeptides that bound with high affinity to a carbonic anhydrase target [J. Am. Chem. Soc., 124, 5692 (2002)]. More recently, they found that when this hydrolysis process is linked with several cycles of dipeptide synthesis, the best binding dipeptide is concentrated in greater than 100-fold excess over more weakly binding dipeptides. This selectivity is more than an order of magnitude greater than would have been achieved in a conventional DCC study, the researchers say.

COMBI CATALYSIS. Combinatorial chemistry is also seeing increasing use of late for discovering and optimizing catalysts. Last March, researchers from Dow Chemical and Symyx Technologies, Santa Clara, Calif., reported having discovered a new class of single-site catalysts for olefin polymerization by high-throughput screening methods [J. Am. Chem. Soc., 125, 4306 (2003); C&EN, April 7, page 10]. They synthesized a library of hafnium and zirconium complexes and screened them by rapidly carrying out 384 polymerization experiments. The work uncovered a number of polyolefin catalysts, one or more of which are being considered for commercialization.

Chemistry professor Scott J. Miller and coworkers at Boston College recently used library screening to find synthetic peptides that catalyze a range of highly selective enantioselective reactions such as phosphorylations, acyl transfers, and conjugate additions [J. Am. Chem. Soc., 124, 11653 (2002); C&EN, Oct. 8, 2001, page 9]. They've used combinatorial means to optimize the activities and selectivities of some of these peptide catalysts. Most recently, they extended the scope of the peptide catalysts to accelerating enantioselective C–C bond-forming reactions as well [Org. Lett., 5, 3741 (2003)].

Ganesan's group at the University of Southampton identified Lewis acid catalysts for a set of highly efficient one-pot Pictet-Spengler reactions by parallel screening. The researchers then used the reactions to produce libraries of diverse cyclized tetrahydro--carbolines [Chem. Commun., 2003, 916]. "Our latest reagent, benzyltriethylammonium tetrachloroaluminate, is an exceptional catalyst [that] may well replace the popular AlCl₃ or 1-butyl-3-methylimidazolium AlCl₄ for many applications," Ganesan says. "We have an inexpensive, polymer-supported version that can be recycled."

SYNTHESIS. The development of novel methods for combinatorial synthesis continues to be of central importance in combinatorial research. For example, new linkers for solid-phase synthesis are being devised to improve the ways that starting materials can be attached to solid supports and products can be released. Ganesan and coworkers recently developed a new tetrafluorobenzenesulfonyl linker "that will have a lot of applications, including traceless cleavage as well as cross-coupling of aryl and vinyl triflates," Ganesan says. Traceless linkers make it possible to cleave products from solid supports without leaving any trace of the functional groups used to attach them to the supports in the first place.

	Full-sized Image

New organic reactions can increase the efficiency and versatility of combinatorial syntheses. Associate professor of chemistry Dalibor Sames and graduate student Bengü Sezen at Columbia University devised a set of new C–H bond functionalization methods to systematically derivatize heterocyclic core structures for diversity synthesis by adding an aryl group to the carbon atom of each targeted C–H bond.

Sames and coworkers recently used this approach to arylate imidazole and 2-phenylimidazole core structures, using different combinations of reaction conditions and aryl donors [J. Am. Chem. Soc., 125, 10580 (2003)]. In the study, they targeted three different C–H bonds selectively in the presence of free amine--a difficult chemical challenge. The team also is expanding the selective arylation method beyond imidazoles to other heterocyclic substrates such as oxazoles and thiazoles [Org. Lett., 5, 3607 (2003)].

Associate professor of chemistry Dennis G. Hall and coworkers at the University of Alberta, Edmonton, are developing new solid-phase synthetic routes to libraries of boronic acid-containing compounds. These compounds are not only useful synthetic intermediates but also have a range of known biomedical applications in their own right, including protease inhibition and carbohydrate recognition and sensing.

Hall's group developed an N,N-diethanolaminomethyl polystyrene (DEAM-PS) resin that uses a rare example of a water-cleavable linker. The resin makes possible "the first general solid-phase approach for the derivatization of functionalized boronic acids," Hall says [J. Org. Chem., 67, 3 (2002)]. DEAM-PS-supported boronic acids can be used for resin-to-resin transfer reactions, which simplify library syntheses by eliminating intermediate steps. They're also applicable to phase-switch chemistry, which makes it possible to carry out homogeneous reactions without workup and purification, thus combining advantages of both solution-phase and solid-phase synthesis. DEAM-PS resin is available commercially.

Another method that combines advantages of solution- and solid-phase syntheses is the precipiton technology devised by chemistry professor Craig S. Wilcox and coworkers in the department of chemistry and the Combinatorial Chemistry Center of the University of Pittsburgh [Angew. Chem. Int. Ed., 40, 1875 (2001); C&EN, June 4, 2001, page 49]. Precipitons are soluble molecular fragments that are added to starting materials for solution-phase synthesis. After synthesis is complete, the precipitons are structurally isomerized to an insoluble form, so products can be removed from solution as precipitates.

The precipiton technology thus combines the ease and versatility of solution-phase synthesis with the easy separability of solid-phase workup, and it can greatly reduce the amount of solvent used in chemical synthesis. Wilcox and coworkers recently improved it by making precipitons recyclable as well. "The recycling is quite efficient now and, combined with savings in solvent use and time, will enhance the attractiveness of this new method of separation," Wilcox says.

New multicomponent reactions are being developed by several groups. MCRs are processes in which three or more starting materials are combined in a single reaction vessel ("one pot") to generate a product that incorporates substantial portions of all the reactants. MCRs are of growing importance in combinatorial chemistry because of their high productivity and their potential to generate libraries of high complexity and high diversity using only single-step reactions.

Hall's group developed an aza[4+2]/allylboration MCR as a stereocontrolled route to polysubstituted piperidines [Chem.--Eur. J., 9, 466 (2003)]. The reaction generates compounds with several stereocenters, it has a wide substrate scope, and it incorporates up to four elements of diversity. "The reaction is being optimized on solid support for generating parallel libraries," Hall tells C&EN. "We are also using this type of reaction in target-oriented synthesis" [J. Am. Chem. Soc., 125, 9308 (2003)].

"The development of novel MCRs is indeed a hot spot in past years and has attracted attention of more and more synthetic chemists," says Jieping Zhu, director of research at the CNRS Institute of the Chemistry of Natural Products, in Gif-sur-Yvette, France. "However, the importance of MCRs has been evident for quite a while. In fact, nifedipine, commercialized in the 1970s, has been synthesized by the Hantzsch reaction, an MCR discovered in 1882."

According to Zhu, a noteworthy MCR advance that's much more recent is the development of enantioselective MCRs. Groups that have led the way in this area, he says, include those of Vice President of Chemistry Alexander Dömling at Morphochem AG, Martinsried, Germany; chemistry professor Carlos F. Barbas III at Scripps Research Institute; associate professor Benjamin List at Max Planck Institute for Coal Research, Mülheim an der Ruhr, Germany; and chemistry professor Scott E. Denmark at the University of Illinois, Urbana-Champaign.

At the ACS ProSpectives conference, Zhu discussed the "substrate design" approach to MCRs--designing substrates with appropriate functional groups, so when the substrates are mixed together they will react in a highly ordered and productive fashion to produce an interesting scaffold in high yield. This approach contrasts with target-oriented synthesis, in which a natural product or other target compound is retrosynthetically disconnected into smaller and easily available synthons, and where the initial focus is thus more on the product than on the substrates. Zhu and coworkers recently used the substrate design strategy to develop multicomponent syntheses of a range of structurally distinct macrocycles and polyheterocycles [Angew. Chem. Int. Ed., 41, 4291 (2002); C&EN, Nov. 18, 2002, page 16].

Zhu and Hugues Bienaymé, founder and chief executive officer of Chrysalon, Villeurbanne, France, are currently editing a book on MCRs, to be published by Wiley-VCH. "It will be the first book dedicated to this important research field and should appear at the beginning of 2004," Zhu says. It will cover well-known, newly developed, and emerging MCRs and their use in drug discovery, natural product synthesis, and combinatorial chemistry.

ONE-POT PROCESS Aza[4+2]/allylboration multi-component reaction, devised by Hall and coworkers, combines several simple starting materials with multiple points of diversity (R1 to R4) into a set of densely functionalized heterocyclic structures.

CREATIVE LEAPS. "Combinatorial chemistry is no longer just about making large libraries," Ganesan notes. "Besides the earlier emphasis on drug discovery, a lot of current attention is now concentrated on the chemistry itself and applying combinatorial principles to discovering new reactions and synthetic methodologies. Indirectly, this will have a tremendous effect on drug discovery, as increasing the repertoire of reliable organic reactions will increase the universe of compounds readily accessible by medicinal chemists."

"This stuff is coming along," said Stanford University chemistry professor Paul A. Wender at the ProSpectives conference, speaking of combichem in general. He said he knows of no tool that has instantly produced the best science. "All tools start off being overhyped, they mature, and then they reach levels of impact that go beyond whatever we imagined at the outset.

"So yes, there was a lot of excitement at the outset, and maybe some of it overshot the mark," he said. "But combinatorial chemistry is evolving into a fairly sophisticated science that's really taking on hard problems and solving those problems.

"I'm sure many of you have examples of combinatorial compounds that are preclinical or clinical," Wender continued. "There's no question about it being useful. There's no question about it being with us as long as chemistry is with us. The only issue in my mind is: What can we do next? Where are the next creative leaps going to come from in this area? Let's get on with that!"