Drug discovery resources in academia and industry are not used efficiently, to the detriment of industry and society. Duplication could be reduced, and productivity could be increased, by performing basic biology and clinical proofs of concept within open access industry-academia partnerships. Chemical biologists could play a central role in this effort.
There is a truism among pharmaceutical scientists that the only way to be confident that a protein is a suitable target for therapeutic intervention is in retrospect—after a successful drug has been developed. Although in many cases laboratory studies have successfully pointed to new therapeutic opportunities, experienced drug discoverers can also wax poetic about the many times that outstanding preclinical science encompassing cell-based experiments, RNA interference studies, mouse knockouts, animal models and localization studies in animal models and patient disease tissue led to conclusions about disease mechanisms that were subsequently invalidated in clinical trials. Some notable examples include the hypotheses, pursued in each case by multiple companies in parallel, that (i) targeting the matrix metalloproteases1 and farnesyltransferase2 inhibitors would provide therapeutic benefit for cancer, (ii) targeting the cholesterol ester transfer protein would benefit people with cardiovascular disease3 and (iii) neurokinin NK1 agonists would provide therapeutic benefit in pain4 (Box 1). In each of these cases, the failures in late-stage clinical trials resulted in enormous cost to each company and to the industry in general, while also dashing the hopes of countless patients.
We use these examples to highlight two of the major problems that confront drug discovery. The first problem is scientific: drug discovery is still confounded by our relatively poor understanding of disease mechanisms, particularly in humans. Outstanding basic and clinical science cannot accurately predict the outcome of clinical trials, where the greatest costs are incurred. The second problem is structural. The economic system that drives the drug discovery sector rewards innovative new medicines and encourages competition. On the positive side, this model ensures that every new drug target is pursued by many companies in parallel and that the overall chances of success are increased. However, the race for the 'first-in-class' drug also has negative consequences. First, it engenders secrecy. Academic and industrial drug discoverers are naturally reluctant to share fundamental scientific information about their preclinical drug discovery programs or their clinical trials, even while acknowledging that their collective success is hindered by a lack of scientific understanding. Second, by encouraging duplication, the economic cost of failure in clinical trials is amplified. When a new disease hypothesis is invalidated in the clinic—as is most often the case—all companies with parallel clinical programs lose. These issues have been widely acknowledged by academia, industry and governments, yet it has proven difficult to find a viable solution, even faced with the harsh reality that drug discovery has steadily declined in its productivity over the past 30 years5. With no alternative, drug discoverers have little choice but to continue to develop medicines for targets and diseases that are inadequately understood (Box 2), to pursue drug discovery programs in competition and to maintain the proprietary nature of their drug development programs.
Change is required—but what to change, and how? There is agreement that the failure rate in clinical trials, particularly those that target new processes, poses the most serious economic burden on drug discovery. It is also commonly accepted that failure rates would decrease if pioneer drug targets were validated with better proofs of concept in the laboratory and in humans before the launch of late-stage clinical trials. In acknowledgment of these facts, most academics now realize that validating a drug target requires more than gene expression studies or experiments in animal models, and companies are now requiring clinical proof of concept as a decision step in their drug development strategy. The problem is that these studies are done inefficiently. Academic researchers are often exploring the therapeutic relevance of new drug targets without access to the best pharmacological inhibitors, which are the most useful and informative research tools from a drug discovery perspective. Industrial scientists perform clinical proofs of concept for pioneer drug targets in parallel and in secret, with no collective learning and with considerable duplication of effort.
To increase the number of pioneer targets with proof of concept in the laboratory and in the clinic, we argue that drug target validation both in the laboratory and in people would be more efficiently performed in a precompetitive environment. Pharmaceutical chemists and academic scientists could combine their knowledge and experience to produce pharmacological research tools, which would be made available to the research community to accelerate the understanding of the role of the target in disease. Clinical proof-of-concept studies for selected pioneer targets would be considered precompetitive; they would be performed by an open access consortium of academics and industry and financed by both sectors. By reducing duplication, the same total resources would generate proofs of concept for a larger and more diverse range of targets. Industry could then focus drug development resources on a greater number of clinically validated targets, improving the likelihood of a positive outcome in phase 3 trials. Ultimately open access science will make industry more profitable and lead to the development of more medicines. In this article, we (i) describe how chemical biologists are required to achieve these goals, (ii) share our nascent experiences with a new model for precompetitive, open access chemical biology and (iii) provide some ideas about how to organize a clinical proof-of-concept consortium.
Open access chemical biology
Potent, selective and cell-permeable inhibitors of protein function (“chemical probes”) are valued reagents in both fundamental and applied biological research, and they are essential for preclinical target validation in academic and industrial laboratories6,7. However, chemical probes are not widely available because they are difficult to produce without access to skilled medicinal chemists; they are also frequently targeted to the relatively few proteins that have already been the focus of industrial drug discovery efforts and are often encumbered by intellectual property and restrictive material transfer agreements. Moreover, many of the probes currently available are inadequately characterized and nonselective, and thus are used inappropriately by the research community. One solution to this problem might be for industry to use their medicinal chemistry expertise to provide chemical probes for all potential drug targets. However, the decreasing productivity of industry requires them to apply more effort on later-stage drug development and to move away from target discovery. The situation has created a paradox: industry is increasingly dependent on academia to discover and validate new targets, yet target validation is optimally done with the use of well-characterized chemical probes, whose derivation is best done in industry.
As one approach to breaking this impasse, we created a new type of partnership in which industry and academia are collaborating to generate chemical probes for “pioneer” targets—those targets for which the biological understanding is poor and clinical validation is lacking—focusing first on proteins involved in the regulation of epigenetic signaling. To ensure that the chemical probes will be used immediately and with maximal benefit, they will be treated as precompetitive reagents and made available to all researchers without restriction on use. To enable the biomedical research community to better interpret their experiments, the collaboration is also committing to provide selectivity and specificity profiles that greatly exceed any scientific standard adopted by academia or industry. Taken together, we expect that the chemical probes will enable a scale and depth of experiments on epigenetic signaling by both academic and industrial investigators that could not otherwise be achieved (Supplementary Table 1 online).
The partnership—which comprises the Structural Genomics Consortium (SGC), the Universities of Oxford and Toronto, the US National Institutes of Health (NIH) Chemical Genomics Center, GlaxoSmithKline and a network of academic collaborators—is taking a systematic approach to generating the chemical probes. Rather than selecting specific 'therapeutically relevant' proteins at the outset, all the important protein families involved in modifying and recognizing histone marks being studied in a systematic way8. This research builds on the output of the SGC, which has produced most of these human proteins in purified form and has determined the three-dimensional structures of many9. For each protein family, ligand-mimetic chemical libraries are designed and synthesized by medicinal chemists using structure-based approaches. The libraries are screened at each of the collaborating groups, using a range of approaches. Iterative improvement of potency and selectivity will exploit the family-wide availability of the purified proteins at the SGC and the growing level of structural understanding of the protein families and their interactions with ligands. The overall progress is being overseen by a committee of scientists having expertise in epigenetics and chromatin biology (Mark Bedford, MD Anderson Cancer Center; Roger Kornberg, Stanford University; Gioacchino Natoli, European Institute of Oncology; Chris Wynder, McMaster University; Ming Ming Zhou, Mount Sinai School of Medicine); this committee will also serve as a conduit for information transfer to the broader community. Promising molecules will be characterized in cell-based assays, and the results of these experiments will guide the design of the final chemical probe, which will be released to the public once it has met predefined criteria for quality and utility (for example, potency, selectivity and cellular activity; Table 1).
In designing the framework for the partnership, the group took heed of the following key lessons discovered over the past 15 years of research in industry:
Chemical tractability is an important, if not the most important, consideration when prioritizing potential targets for drug discovery. Arguably, the most efficient approach to identifying selective inhibitors and innovative drug targets is to focus on chemically tractable protein families.
Smart experimental strategies often beat brute force. Knowledge-based chemical library design and the use of three-dimensional structural information can expedite the development of probes (these strategies are three to ten times faster than relying solely on high-throughput screening).
Synthetic ligands are powerful tools for discovery of new biology only if they are well characterized, potent and selective for one or a subset of targets within a protein family, and only if they are used in bioassays that are able to distinguish physiology from phenomenology.
Involvement of the scientific community is critical. Everyone stands to benefit from advances in our knowledge of the target and its role in human biology—and rarely does one group or organization have all the necessary resources or capabilities to validate a wide range of targets.
And will the chemical probes be useful? Our experiences with orphan nuclear receptors and protein kinases show that high-quality, well-characterized chemical probes, if made freely available, would not only inform drug discovery, but would also have an enormous impact on our basic understanding of human biology. In the late 1990s, Glaxo Wellcome (and its successor GlaxoSmithKline) released a select number of chemical probes for orphan nuclear receptors to academic researchers. These compounds—which were released initially under material transfer agreements but later with no restriction on use—fueled research, functional validation and drug development for disorders of lipid, sterol and bile acid metabolism10,11,12. When assessed using bibliometric methods, the scientific impact of each chemical probe is on par with that of top academic researchers (Table 2). Likewise, the field of kinase signaling was revolutionized by the discovery of the small-molecule, cell-active inhibitor staurosporine13. The subsequent discovery, publication and wide availability of selective kinase inhibitors has enabled many chemogenomic approaches aimed at elucidating their roles in biology and human diseases, as well as the development of new cancer therapeutics such as imatinib, getfitinib, lapatinib and erlotinib. Staurosporine is also an excellent example of the need for better chemical probes. Staurosporine is a nonselective pan-kinase inhibitor, yet there are over 8,000 publications that use staurosporine as a specific modulator of cell signaling, including over 1,200 in the past 4 years.
Our goal in providing these research tools to the community without restriction on use is to spur research into epigenetics and chromatin biology, and to generate hypotheses about potential therapeutic opportunities. However, the success of the program and the continued involvement of industry both rely on the scientific community's willingness to use the chemical probes properly and share their data without restriction on use. We therefore urge the community to adopt these tenets.
Open access clinical trials
The availability of chemical probes for proteins implicated in chromatin biology is expected to generate both new hypotheses about function and new therapeutic opportunities. However, as described previously, the crucial experiment from a drug discovery perspective is to test whether these and other hypotheses hold true in humans, and this can only be achieved with careful and well-controlled proof-of-concept clinical trials.
In the current division of labor between the public and private sectors, proof-of-concept studies are usually carried out by industry. This has some logic because (i) the costs—which include the preclinical development of the inhibitor and the clinical studies themselves—are deemed too high for the public sector; (ii) the relevant expertise is largely found in industry and (iii) any eventual commercial rewards are given to the risk-taking company. However, the model has deficiencies, as ultimately evidenced by the diminishing productivity of the pharmaceutical sector. The fundamental problem is that industry collectively focuses too many resources on proof-of-concept studies for too few targets, and the studies are done in a proprietary way, with little collective learning. Further, because one 'secret' failure in proof of concept is never enough to dissuade others, these studies encumber the limited resources of industry for years, thereby limiting the ability of industry to pursue new and potentially relevant drug targets.
Our aim here is to propose a mechanism to significantly increase the number of clinically validated new drug targets for the same cost. To accomplish this, we argue that clinical proof-of-concept studies for selected targets should no longer be considered as a step on the path to commercialization, but rather as a precompetitive scientific experiment whose output can therefore be made available to all, without restriction on use (Supplementary Table 2 online). The dissociation of clinical proof of concept from commercialization is not only realistic—given that very few first-in-class compounds for new targets are ultimately developed into the first marketed drug—but also strategic because it would enable alternate funding schemes, facilitate cross-sector collaborations, expand the number of scientists thinking about the problems and allow decisions to be made with a greater emphasis on science and less on the perceived size of the market.
We propose that academia and pharmaceutical companies should form a precompetitive consortium whose objective is to develop clinical probes to achieve proof of efficacy in humans for a wide range of potential targets. The consortium, which would be financed by all stakeholders, would use established ethical and regulatory guidelines and seek out the best ideas, the best technologies and the most advanced biomarkers to ensure that the clinical probes (i) are potent and selective; (ii) meet appropriate pharmacodynamic and pharmacokinetic endpoints; (iii) are tested in the most appropriate diseases; (iv) have safety profiles that allow appropriate clinical exposures to fully test the mechanism; and (v) inhibit the function of the target in the tissue of relevance in a concentration-dependent manner. The consortium would also ensure that the appropriate scientific clinical endpoints are achieved, and would commit to publishing all of the data as early as possible and with no restriction on use. To manage costs and to achieve maximal benefit, it will be necessary to select the targets and the indications carefully; the consortium must select projects that can achieve meaningful clinical proof of concept after a phase 2a study. Although we fully expect that the clinical probes developed by the consortium will fail as often as any developed by industry, the pooling of resources reduces duplication and allows the interrogation of a larger number of pioneer targets.
Dissociating the development of clinical probes from commercialization also obviates the need to seek patent protection on the molecule or to ensure that the molecule is novel, from an intellectual property perspective (Box 3). This is a tremendous advantage of an open-access approach because it avoids lengthy negotiations over interinstitutional material transfer and collaboration agreements, and it makes possible collaborations between scientists in different institutions and countries. The absence of a patent position does, however, raise a legitimate question: what happens to a clinical probe that appears to have the appropriate characteristics to progress into phase 2b and phase 3 clinical trials? We believe that there are many ways that promising molecules could be developed using other financial models. In the absence of a patent position, governments could grant the organization that funds a large clinical study a period of exclusivity in their jurisdiction based on the existing concept of 'data exclusivity' (http://www.who.int/intellectualproperty/topics/ip/en/DataExclusivity_2000.pdf). Alternatively, industry may choose to progress the nonproprietary compound by creating new intellectual property—perhaps using a new formulation of significant therapeutic benefit or by co-administration with proprietary molecules. At the end of the day, we are confident that the creativity of the business community and the needs of the patients will ensure that any compound that looks promising in a proof-of-concept study will be progressed.
Of course, there are many obstacles to be overcome, including how exactly to finance the consortium; how to ensure open access; how to balance the interests of the public and private sectors in selecting the research agenda; how to select the drug targets; how to implement decisions about which probe programs to progress and to stop; how, if possible, to distinguish compound-related from target-related effects; and how to involve regulators in the process. It is also important to learn how to manage the risks of open-access science—namely that the clinical probes will inevitably be used to carry out low-quality experiments and muddy the scientific scene.
All this said, these obstacles are perhaps insignificant when viewed in light of the current modus operandi in which commercially driven clinical trials fall like dominos in the clinic—to the detriment of each company, to the detriment of the patients and with relatively little communal learning. This approach has a potential social benefit as well: making the drug discovery process more transparent to the public may increase public awareness of the complexity of drug development and lower public mistrust of the pharmaceutical industry.
In summary, the development of new medicines is being hindered by the way in which academia and industry advance innovative targets. By generating freely available chemical and clinical probes and performing open-access science, the overall system will produce a wider range of clinically validated targets for the same total resource. This is arguably the most effective way to spur the development of treatments for unmet needs.
Note: Supplementary information is available on the Nature Chemical Biology website.
Fingleton, B. Semin. Cell Dev. Biol. 19, 61–68 (2008).
Appels, N.M., Beijnen, J.H. & Schellens, J.H. Oncologist 10, 565–578 (2005).
Neeli, H. & Rader, D.J. Cardiol. Clin. 26, 537–546 (2008).
Boyce, S. & Hill, R.G. in Proceedings of the 9th World Congress on Pain (eds. Devor, M. et al.) 313–324 (International Association for the Study of Pain Press, Vienna, 2000).
Garnier, J.P. Harv. Bus. Rev. 86, 68–70, 72–76, 128 (2008).
Knight, Z.A. & Shokat, K.M. Cell 128, 425–430 (2007).
Newman, R.H. & Zhang, J. Nat. Chem. Biol. 4, 382–386 (2008).
Weigelt, J., McBroom-Cerajewski, L.D., Schapira, M., Zhao, Y. & Arrowmsmith, C.H. Curr. Opin. Chem. Biol. 12, 32–39 (2008).
Edwards, A. Annu. Rev. Biochem. 78, 541–568 (2009).
Kliewer, S.A., Lehmann, J.M. & Willson, T.M. Science 284, 757–760 (1999).
Chawla, A., Repa, J.J., Evans, R.M. & Mangelsdorf, D.J. Science 294, 1866–1870 (2001).
Willson, T.M. & Moore, J.T. Mol. Endocrinol. 16, 1135–1144 (2002).
Tamaoki, T. et al. Biochem. Biophys. Res. Commun. 135, 397–402 (1986).
Hirsch, J.E. Proc. Natl. Acad. Sci. USA 102, 16569–16572 (2005).
Menkes, C.J. et al. J. Rheumatol. 20, 714–717 (1993).
Marshall, K.W., Chiu, B. & Inman, R.D. Arthritis Rheum. 33, 87–90 (1990).
Babenko, V.V. et al. Eur. J. Pain 3, 93–102 (1999).
Jensen, K., Tuxen, C., Pedersen-Bjergaard, U. & Jansen, I. Cephalalgia 11, 175–182 (1991).
Pedersen-Bjergaard, U. et al. Peptides 10, 1147–1152 (1989).
Von Euler, U.S. in Neurotransmitters in Action (ed. Bousfield, D.) 143–150 (Elsevier Biomedical Press, Amsterdam, 1985).
Salt, T.E. & Hill, R.G. Neuroscience 10, 1083–1103 (1983).
Mantyh, P.W. et al. Science 278, 275–279 (1997).
Woolf, C.J. & Mannion, R.J. Lancet 353, 1959–1964 (1999).
Rupniak, N.M.J. & Hill, R.G. in Novel Aspects of Pain Management: Opioids and Beyond (eds. Sawynok, J. & Cowan, A.) 135–155 (Wiley-Liss, New York, 1999).
Gunthorpe, M.J. & Chizh, B. Drug Discov. Today 14, 56–67 (2009).
Gharat, L.A. & Szallasi, A. Expert Opin. Ther. Pat. 18, 159–209 (2008).
The authors thank C. Arrowsmith, I. Benjamin, D. Dodds, S. Frye, B. Greenberg, J. Weigelt, H. Widmer, T. Yamada and R. Young for comments on the paper. A.M.E. and C.B. are supported by the Structural Genomics Consortium, which is a registered charity (number 1097737) that receives funds from the Canadian Institutes for Health Research, the Canadian Foundation for Innovation, Genome Canada through the Ontario Genomics Institute, GlaxoSmithKline, Karolinska Institutet, the Knut and Alice Wallenberg Foundation, the Ontario Innovation Trust, the Ontario Ministry for Research and Innovation, Merck & Co., Inc., the Novartis Research Foundation, the Swedish Agency for Innovation Systems, the Swedish Foundation for Strategic Research and the Wellcome Trust.
About this article
Cite this article
Edwards, A., Bountra, C., Kerr, D. et al. Open access chemical and clinical probes to support drug discovery. Nat Chem Biol 5, 436–440 (2009). https://doi.org/10.1038/nchembio0709-436
A responsibility to commercialize? Tracing academic researchers’ evolving engagement with the commercialization of biomedical research
Journal of Responsible Innovation (2019)
Probabilistic modeling of personalized drug combinations from integrated chemical screen and molecular data in sarcoma
BMC Cancer (2019)
Gates Open Research (2019)
Gates Open Research (2019)
MNI Open Research (2018)