Modern drug discovery efforts are exploiting at least three core technologies aimed at increasing the efficiency of finding drug leads: genomics, high-throughput screening, and combinatorial chemistry. Research aimed at deciphering the human genome is rapidly multiplying the number of known disease targets. Screening methods using biological assays can quickly show if a compound is a "hit"; that is, if it has activity against a target.
And combinatorial chemistry methods can produce and help optimize the compounds used in screening. Many pharmaceutical companies view this hot area as technology they must have to compete.
In the past, most drugs were discovered by screening collections of compounds to find a random hit. Companies with larger collections or libraries have had an empirical advantage. Although more compounds still can be better, especially with the arrival of high-throughput robotic screening, an expanding knowledge of the molecular biology of disease is shaping the design of these libraries.
But speed - which translates into time, money, and an ability to compete - is a key concern in drug discovery. Laborious synthesis methods in which a chemist makes one compound at a time cannot keep up with the desired pace. Combinatorial chemistry - broadly defined by many companies as the generation of numerous organic compounds through rapid simultaneous, parallel, or automated synthesis - is changing how chemists create chemical libraries and is expected to change the speed at which drugs are found.
"In the past, the chemist very rightly took pride in having a compound that was very pure because it was expected to have to be active in vitro and to have to go into in vivo models," explains John L. LaMattina, vice president for U.S. discovery operations at Pfizer's Groton, Conn.-based central research organization. "In running [the new high-throughput] screens, you don't care if there are impurities. As long as you generate some activity, you can figure out a lead structure." Screening assays also require only small quantities of material.
The bottleneck in drug discovery no longer is biology, but rather the number of compounds available, believes LaMattina. Even the large pharmaceutical company libraries, traditionally numbering in the hundreds of thousands of compounds, have become insufficient for screening. New sources of novel compounds are needed.
With these realizations, chemists have surmised that small-molecule synthesis can take advantage of automation and robotics, as long as they can have good faith that the mixture being tested consists predominantly of the structure they aimed to synthesize, says LaMattina. "This was the logic that got us to say relatively quickly, 'Let's do chemistry nontraditionally,' " he notes. So, instead of making one compound per week, 100 can be produced in a day.
Pfizer is using in-house combinatorial chemistry to enrich its chemical libraries and as an element of its exploratory medicinals group. The company has gone to outside sources once, setting up a $5 million collaboration with the U.K.-based chiral chemistry company Oxford Asymmetry in March 1995. In turn, Oxford Asymmetry created a wholly owned subsidiary, Oxford Diversity, dedicated to combinatorial chemistry.
In October 1995, Sepracor, another chiral and pharmaceutical chemistry company, created a combinatorial chemistry subsidiary, Versicor. The Marlborough, Mass.-based Versicor is leveraging Sepracor's existing infrastructure for drug discovery and development and its proprietary chiral chemistries. The subsidiary has been working to develop novel, druglike heterocyclic compound libraries.
Other pharmaceutical companies also have stepped up the pace at which they are accessing combinatorial chemistry technologies. Most major drug producers have set up alliances with a new generation of combinatorial chemistry-based drug discovery companies. The interest from pharmaceutical firms, as demonstrated by the more than $500 million pledged to date in R&D alliances, clearly is directed toward small-molecule synthesis. At least another $750 million has been invested in genomic companies to find disease targets (C&EN, Dec. 4, 1995, page 18).
First-generation combinatorial technologies focused largely on peptides and oligonucleotides. Small companies - including Affymax, Gilead Sciences, Houghten Pharmaceuticals, Isis Pharmaceuticals, NeXstar, Protein Engineering, Selectide, and Sphinx Pharmaceuticals - opened their doors in the 1980s when automated DNA and amino acid synthesizers and sequencers were becoming established technologies to produce and identify compounds.
By late 1994, Eli Lilly had acquired Sphinx for $80 million. Marion Merrell Dow then spent $58 million to buy Selectide in early 1995. In March 1995, Glaxo paid the premium price of $533 million to take over Affymax, one of the first, and most technologically sophisticated, combinatorial chemistry companies. Meanwhile, many pharmaceutical producers started putting together in-house efforts.
Most of the major pharmaceutical companies have been developing combinatorial or automated synthesis methods during the 1990s, LaMattina believes. He comments that "part of the reason it maybe hasn't been well advertised is that many companies look at these as trade secrets and not something that is really patentable."
Still, a second generation of combinatorial chemistry companies has emerged with proprietary technologies focused largely on small-molecule chemistries. "It's technology that is much more widely accepted, at least by the major pharmaceutical companies, than most new technologies," says David J. Ecker, vice president of Isis Pharmaceuticals, Carlsbad, Calif. "The reason is that it's not that much different from mainstream drug discovery in the pharmaceutical industry for 100 years.
"It's just a technical advance that allows you to do things much more quickly, but it's not breaking away that much from a tried-and-true approach," adds Ecker, who also is managing director of Isis's combinatorial program. "In that regard, it seems to require much less justification than something like gene therapy, something that's new and never existed before and people don't know what all the hurdles are going to be."
While continuing with its antisense oligonucleotide programs, Isis has devoted a few years to developing nonpeptide, nonoligonucleotide small-molecule syntheses. The company's basic approach is to use a series of novel chemical "scaffolds" that can be combined and attached with multiple functional groups to create diverse compound libraries.
Isis specifically avoids working with structures of existing pharmacophores. "The most value will come from novel chemical structures applied to unexplored biological targets," stresses Ecker. "They then won't be 'me too' compounds [because they will be] structurally different from the things that exist today and because they are being leveraged at new targets where we don't have a lot of drugs already."
Analogs of benzodiazepines, the family of heterocyclic compounds that includes the major tranquilizer drug Valium, are frequent targets of combinatorial programs. The logic is simple: Take a known low molecular weight drug that fundamentally has good structural and desired pharmacological properties and create new analogs through combinatorial chemistry. Although making analogs of known structures runs the risk of creating me too drugs, the same idea can be used in optimizing the activities of novel drug leads.
"Getting the leads is what is most important and is what really becomes patentable," explains LaMattina. "We're not going to patent a broad library, per se, because its utility would be very vague." In the past, a drug company did its best to optimize its lead and then protect its intellectual property by individually synthesizing as many diverse analogs as possible within time and cost constraints.
However, after finding a lead from screening a large combinatorial library, one can "in theory apply this chemistry to patent protection [by creating] thousands of compounds around an active structure and making it very difficult for competition to break through," LaMattina says.
To increase the chances of finding leads, some combinatorial chemistry companies have taken the approach of producing very large libraries. Often, combinatorial chemistry gives small companies chemical libraries that are on the same order as those once the domain of only the largest drug firms.
With these libraries and their proprietary technologies, small companies can leverage lucrative deals that include large up-front and milestone payments, and longer term royalties. The deals, in turn, give the small companies access to biological targets and the pharmaceutical infrastructure necessary to take drug candidates through development, clinical testing, regulatory approval, and marketing.
To generate revenues, ArQule, Medford, Mass., has drug discovery partnerships with Solvay of Belgium, Illinois-based Abbott Laboratories, and Pharmacia Biotech of Sweden. After screening ArQule's libraries, collaborative efforts with Abbott and Solvay will work with more directed arrays of compounds to refine the development of specific product candidates.
ArQule uses what its calls "modular building-block" chemistries - or groups of reactive monomers - combined with an automated system for synthesis. According to the company, its solid- and solution-phase chemistries are "robust and flexible" and easily scaled up. One of the company's most recent advances was to produce a small-molecule, peptide mimetic in collaboration with researchers at Brandeis University, Waltham, Mass. (C&EN, July 10, 1995, page 6).
Pharmacopeia also leverages its technology through partnering strategies. The Princeton, N.J.-based firm will license its compound libraries to others or will take a partner's selected disease target and conduct biological screening in its own labs to optimize and deliver a drug lead. In the long term, the company plans to identify its own targets and find drug leads.
"The trade-off is always a matter of risk and reward," says Joseph A. Mollica, chairman and chief executive officer of Pharmacopeia. "The path is not completely linear in terms of investment relative to return." Mollica foresees the company focusing only on the early stage of discovery and development. "We'll demonstrate the hypothesis [of a drug lead] in a relevant animal model and then out-license the compound because there is a lot of investment required before you get to the next big return."
Pharmacopeia has already set up some of the largest deals to date. Under an agreement with Sandoz Pharma, potentially worth $100 million, Pharmacopeia is providing the Swiss drug producer with hundreds of thousands of compounds to screen in several drug areas. Sandoz, in partnership with the San Diego-based venture-capital firm Avalon Ventures, funded the start-up of Pharmacopeia and of Argonaut Technologies, a San Carlos, Calif.-based developer of automated organic synthesis instrumentation and reagents.
In other deals, Pharmacopeia will receive up to $20 million each from Bayer and from Berlex Laboratories for combinatorial compounds and high-throughput screening against partner-supplied targets. And in a $75 million deal with Schering-Plough, the company supplies compounds as well as conducts screens against cancer and asthma targets in its own labs. In all three deals, Pharmacopeia is to receive royalties on resulting products.
Tagging or "split and pool" technologies used by Pharmacopeia and others are considered by some to be "pure" combinatorial chemistry, in contrast to parallel synthesis, which keeps synthesized compounds separate. Pharmacopeia's sophisticated tagging methods are based on work by W. Clark Still of Columbia University and Michael H. Wigler of Cold Spring Harbor Laboratory in Cold Spring Harbor, N.Y. Encoding tags allow for the identification of diverse compounds produced in very large mixtures.
Mollica argues that Pharmacopeia's approach allows for the production of significantly larger, more diverse libraries. Parallel synthesis methods are limited, he believes, by the number of compounds that can be handled or deciphered through analytical methods. Pharmacopeia's chemistry focuses on heterocyclic structures, he adds, which are "the mainstay of the drug industry."
Houghten Pharmaceuticals, one of the first combinatorial chemistry companies, founded by Richard A. Houghten of Scripps Research Institute, has shifted its focus to heterocyclic, small-molecule chemistry in the past two years. One of its early peptide-based combinatorial drug leads is now in Phase II clinical trials. Because of the interest in combinatorial chemistry, the San Diego-based company has set up nine partnerships, says Robert S. Whitehead, Houghten president and CEO. "Longer term, we'd like to focus internally on a selected series of our own discovery programs."
Many other companies see an advantage in developing smaller, more focused libraries, and making larger quantities of each compound. Parallel or robotic synthesis therefore is amenable because smaller libraries do not need to be encoded to be sorted out. Analytical control over the chemistry and in identifying compounds, say many executives, is as or more important than making millions of compounds.
For example, Versicor is "developing chemistries that address biological themes," according to company president Jeremy Goldberg. These themes include compounds that mimic protein structures. "There is plenty of room in small-molecule combinatorial chemistry to address shape space using chiral and stereoselective methods ... in a way that mimics the chirality of targeted enzymes or receptors," he says.
Mimicking the shape of molecular recognition interactions, many in the field believe, will enhance the chance of finding drug leads. Molecumetics, Bellevue, Wash., creates small libraries based on conformationally constrained peptides that mimic protein secondary structures such as &bgr;-turns, &agr;-helices, and &bgr;-sheets. These template libraries are then used to synthesize nonpeptide, small-molecule drug leads.
"The real proof is in the actual molecules that you can find, and so we're concentrating to a large degree on coming up with lead compounds and using those as a source for partnering," says Edward Field, director of business development at Molecumetics. The company currently is collaborating with Bristol-Myers Squibb.
San Diego-based CombiChem, whose technology is also based on work by researchers at Scripps Research Institute, directs its programs toward gene families, says the company's chief operating officer, Peter L. Myers. Genomics will lead to the discovery of multiple receptors and repeated opportunities to sell libraries directed at different gene families. Myers and many others in the field believe that combinatorial chemistry will continue to move toward producing smaller, more directed libraries.
Myers says CombiChem also is developing a "universal signature library" based on the premise that about 10,000 or fewer molecules can represent 'chemical space' - that is, they are rich enough in different chemical and structural features and geometry to be capable of interacting with a whole range of targets. In screening this general library, believers say there is a reasonable chance of finding a hit, even a poor one, to use as an initial lead.
"It's a quite different approach from what I call the brute force approach," says Myers. "Instead of synthesizing millions of compounds, we're trying to design about 10,000 compounds very well." However, there are skeptics within both the academic and corporate combinatorial chemistry communities who question the utility of universal libraries.
ChemBridge Corp., Northbrook, Ill., has just made one such library available commercially. Its DIVERSet96 library of druglike molecules comes from historic synthetic collections of the former Soviet Union's chemical community. Using pharmacophore diversity analysis software developed by Chemical Design Ltd., Chipping Norton, U.K., a subset of the Soviet compound collections was "carefully chosen to cover the maximum diversity of pharmacophores [with] the minimum number of compounds," according to the company.
After a hit is found from a universal library, the idea, says Myers, is to "then build second-generation libraries around the regions of structure-activity space that those weak hits represent, and go on in an iterative fashion like that until eventually you get compounds that are really quite potent." As a business strategy, he suggests that the company can license the proprietary universal library again and again to drug companies. And CombiChem would collaborate on the optimization process to develop drug candidates.
Besides working in the drug discovery area, CombiChem is looking to market instrumentation and software. "There is a sizable market out there," Myers says. "There is not really a decent instrument [available] and people are cobbling together robotic systems." Instrumentation initially will be quite expensive, he believes, but within a few years will come down in price. Eventually, Myers and others believe that individual laboratories probably will be able to afford bench-top machines that can be integrated with other bench-top analytical and screening equipment.
In addition to CombiChem, Argonaut and some of the more traditional laboratory instrumentation companies have begun to enter the automated organic synthesis instrumentation area. Parke-Davis, a division of Warner-Lambert, has created a company, Diversomer Technologies, that plans to commercialize its system. SRI International's David Sarnoff Research Center has created Orchid Biocomputer to collaborate with SmithKline Beecham in developing a miniaturized device for molecular synthesis and screening.
The move to create commercial instrumentation may be part of a larger trend. With instrumentation available and experience with the technology increasing, most in the field believe that it is simply a matter of time - some say two to three years, others five to 10 years - before combinatorial chemistry becomes a core technology at most pharmaceutical and drug discovery companies. But such a move by the industry clouds the outlook for long-term growth or success of small combinatorial chemistry-based companies.
"It's an idea whose time has come, we don't have to convince anybody of the power of the technology," says Mollica. "But there is a viable strategy to being an independent company and to grow and prosper based on the strength of our technology." Larger companies will continue to make acquisitions, to invest heavily to put the technology in place, or to set up alliances, he and other executives believe. Others, however, suggest that libraries and instrumentation soon may become "commodities."
"Is it a long-term business? The longer term profitability is really in the therapeutic drug area [and being] a company that continues to drive potential products toward the clinic," adds Barry Toyonaga, president of Carlsbad, Calif.-based Ontogen, which stresses its well-characterized, but modest-sized, library. The company's combinatorial chemistry technology is based in part on work using tagging approaches developed with Robert W. Armstrong of the University of California, Los Angeles.
"We're not in the commodity business," he stresses. "I am not in the game of making us a long-term technology business, because I don't think it is. I think the technology will become more widespread and everyone will have a different flavor of it.
"Obviously, we need to keep the lights on here, so in the short term what we have to sell is technology and compounds for R&D collaborations." Like many other companies, Ontogen hopes eventually to fund internal drug discovery programs. "We've always said that we should run meaningful research programs and, when we get a chance to speak, point to hard biological data, not our shiny machine," Toyonaga says.
This philosophy is evident at other drug discovery companies - such as Arris Pharmaceutical, Ariad Pharmaceuticals, Chiron, 3-Dimensional Pharmaceuticals, and Neurogen - that are combining combinatorial chemistry with structure-based drug design, molecular biology and biological target identification, high-throughput screening, computational chemistry, and other techniques. Almost all these firms have collaborative alliances in which pharmaceutical partners are seeking not only combinatorial chemistry, but also the value brought by other drug discovery approaches.
3-Dimensional Pharmaceuticals' patented DirectedDiversity technology combines computer refinement of combinatorial libraries in conjunction with 3-D structural data. Neurogen is using combinatorial chemistry to find neuropsychiatric drug candidates. It also has generated and licensed an antiobesity drug to Pfizer and is working in collaboration with Schering-Plough. Arris has major deals employing its technology mix with Bayer and Pharmacia, whereas Chiron and Ciba-Geigy are working together.
Ariad has a $40 million collaborative agreement with Hoechst Marion Roussel to develop small-molecule drugs against osteoporosis. The "architect" of Ariad's technology and approach to drug discovery, says Harvey J. Berger, chairman and CEO of Ariad, is Stuart L. Schreiber of Harvard University. Schreiber is one of Ariad's founders and chairman of its scientific advisory committee. Berger uses the term "structure-based combinatorial chemistry" to describe the integration of rational structure-based drug design, molecular biology, and lead compound optimization.
"We're using combinatorial chemistry in every one of our programs, across all of the different drug discovery efforts now and will in the future," says Berger. "Combinatorial chemistry is a tool largely available to everybody. But it's a big investment in people and infrastructure and all of the components, so it's not something you can just go to a catalog or hire a postdoctoral fellow and get it to work."
Once in place, combinatorial chemistry methods are very "learnable" with the right mind-set, say company executives, who are looking for well-trained organic chemists. "We look for creative chemists who can take the technology further and who can develop even more types of chemistries that are applicable," says Pfizer's LaMattina. Solid- and solution-phase reactions that can be used for library building are appearing more often in the scientific literature, he adds.
However, Ontogen's Toyonaga says that it's harder to hire chemists at smaller companies. "There's no history of bench chemists getting rich, whereas every biologist has a friend or colleague who joined a biotechnology company and got rich. Chemists have not had that opportunity," he notes. "This is the first wave of chemistry-based companies that hopefully will become places where chemists can take their talents and really create something both professionally and personally."
In their favor, most combinatorial chemistry companies and other small drug discovery firms have experienced management teams culled from the pharmaceutical and biotechnology industries. Many also have prominent scientific advisory board members, chief scientists, and scientific founders who have contributed to the companies' core technologies.
Ultimately, the bottom line will be whether the technology enables companies to discover and develop drugs more rapidly. "Everybody has combinatorial chemistry on their [company presentation] slides, but there are huge differences," says Berger. "The critical question is who actually has it up and running, is making molecules, and is getting it to work. How well you do it, how effectively and creatively you use it, and how you actually apply it is going to separate the good companies from the bad."
Although there is disagreement about which approach will create the most value - that is, increasing the numbers of drugs that reach clinical trials - there is a consensus that the technology is still too young to have generated clear success stories. However, many point to promising leads in early development that have been generated in months rather than years. And if problems arise with an initial drug lead, companies stress that their libraries can serve as sources of follow-up compounds.
According to LaMattina, Pfizer has a compound in early clinical trials that researchers there wouldn't have discovered without high-speed chemistry. "It wasn't discovered by building a 500,000-compound library," he explains, "but we used this technology to make about 1,000 compounds in relatively short time around one specific lead.
"Interestingly enough, we didn't make the big chemical breakthrough until we made about compound 900," he says. "And in the old days, I can tell you, we would have made about 300 compounds and if we hadn't improved the activity by then we probably would have dropped it."
[ACS Home Page]
[ACS Publications Division Page]