Too Many Roads Not Taken Most Protein Research Focuses on Those Known Before the Human Genome Was Mapped

Home , Nature

COMMeNT antHropoloGy Call for Genetics Reviewed: two policy Sanitation, not obituary Jack Oliver, key unity in the science of primers on personal vaccination, is most player in proof of plate human beings p.166 genomics p.169 important in Haiti p.175 tectonics, remembered p.176 urton urton b on by Jonathan Jonathan by on I Illustrat

Too many roads not taken Most protein research focuses on those known before the human genome was mapped. Work on the slew discovered since, urge Aled M. Edwards and his colleagues.

hen a draft of the human genome We performed a bibliometric analysis to remain ‘hidden in plain sight’, with scientists was announced in 2000, funders, assess how research activity has altered over proving very reluctant to study them. governments, industry and time for three protein families that are cen- Where there has been a shift in research Wresearchers made grand promises about how tral in disease and drug discovery: kinases, activity, it was often spurred by the emergence genome-based discoveries would revolu- ion channels and nuclear receptors. For all of tools to study a particular protein, not by a tionize science. They promised that it would three, we found very little change in the pat- change in the protein’s perceived importance. transform our understanding of human biol- tern of research activity — which proteins We believe that ensuring high-quality tools ogy and disease, and provide new targets for are associated with the highest number of are developed for all the proteins discovered drug discovery. Yet more than 75% of protein publications — over may be all that is needed to drive research into research still focuses on the 10% of proteins nature.com the past 20 years1. the unstudied parts of the human genome — that were known before the genome was Protein mapping Even those proteins even within funding and peer-review systems mapped — even though many more have gains a human focus: that have been directly that are inherently conservative. been genetically linked to disease. go.nature.com/vbqctf associated with disease We searched for mention of every human

kinase, ion channel and nuclear receptor in activity in 2009 was focused on just three. required to generate and characterize such either the title, abstract, keywords or ‘MeSH’ These three were also ‘top of the nuclear tools, they are currently available for only a terms (used for indexing articles in Medline receptor charts’ in the 1990s. handful of well-studied proteins. and PubMed) in the almost 20 million papers Why the reluctance to work on the Our analysis of publication patterns for published between 1950 and 2009. We dis- unknown? As the Nobel-prizewinning bio- the human nuclear hormone receptor fam- covered that for all three classes of protein, chemist Roger Kornberg put it, scientists are ily suggests that making such tools readily the same small fraction of family members wont to “fondle their problems”: they have available for all proteins could dramatically have remained ‘the favourites’ for nearly 20 a natural tendency to dig deeper into their shift the balance in biomedical research. years (see ‘Fondling our problems’). areas of expertise. Plus, funding and peer- For instance, the human genome encodes review systems are risk-averse; funders and wake-up call more than 500 protein kinases, of which hun- reviewers alike are less willing to support Nuclear receptors are transcription factors dreds have been shown to have genetic links research on unstudied proteins, for which it that bind small signalling molecules, such with human diseases. Yet around 65% of the is often harder to explain the rationale and as steroids and hormones. Genetic data now 20,000 kinase papers published in 2009 focused significance. Moreover, the time frames suggest that all the receptors are directly or on the 50 proteins that were the ‘hottest’ associated with academic promotion and indirectly linked to human disease. About in the early 1990s. Similarly, 75% of the training encourage researchers to focus on 30 of the family members were discovered at research activity on nuclear hormone recep- systems that are likely to generate results around the same time in the 1990s, allowing tors in 2009 focused on the 6 receptors — out rapidly, and for which research infrastruc- us to compare publication trends for numer- of the 48 encoded in the genome — that were ture and methods are already available. ous related proteins over time. We know most studied in the mid 1990s (ref. 1). Some funders are developing strategies exactly when the receptors were cloned, to address the conservative nature of peer when genetic links with diseases were estab- biased approacH review. The Wellcome Trust, the largest lished and when research tools (in this case, Although academics may be surprised by the non-governmental funder of biomedi- chemical probes)4,5 became available. magnitude of this research bias, they gen- cal research in the United Kingdom, for When the ‘novel’ nuclear receptors were erally acknowledge its existence. It was first instance, is withdrawing its project grants identified in the 1990s, all the family members identified in kinase research2 in 2008 and last in favour of providing longer-term support were thought to have therapeutic potential. year its effects were demonstrated in kinase to outstanding investigators. And many uni- Interest developed 3 drug discovery . But a common assumption is versities are examining the pitfalls of their “Making most rapidly in those that previous research efforts have preferen- current reward systems. Unfortunately, protein-based that were found to tially identified the most important proteins. institutional systems are ponderously slow have genetic links to research 6–8 The evidence doesn’t support this. to change. So what else can be done? tools readily disease or that had Patterns of gene expression and links To establish a protein’s function, and interesting knockout available must 9 between DNA sequences and breast cancer especially the details of how it works and be a major phenotypes , such as suggest, for instance, that 11 protein kinases its suitability for drug discovery, molecular infertility. However, are key nodes in the signalling pathways biologists draw on an arsenal of tools. For objective in over the next 15 years, underlying the disease. Yet in the 2009 lit- instance, antibodies can help them iden- the decade to research activity refo- erature, one of these kinases, CDC2, received tify where in the body the protein is being come.” cused on a subset of more attention than seven of the others expressed; chemical inhibitors can be used 8 of these receptors. combined, and three received just one men- to block a protein’s activity in human cells From a genomics point of view, these 8 are tion. Likewise, various genetic approaches, and in animal models. These antibodies no more interesting than any of the other including genome-wide association stud- and small molecules also provide a launch 29 with known links to disease. ies, have directly linked 37 of the 48 human pad for the development of new medicines To our knowledge, the only connection nuclear receptor genes to disease. Among by the biotechnology and pharmaceutical among these 8 receptors is that for each there these, more than half of the total research industries. Yet because of the cost and time is a widely available, high-quality chemical probe that either enhances the receptor’s activity or dampens it. In short, where high-quality FONDLING OUR PROBLEMS tools are available (often commercially), there Researchers’ ‘favourite kinases’ have remained the same for decades with a few exceptions (kinases is research activity; where there are no tools, linked to diseases of great interest to industry). there is none (see ‘Tools are telling’). 15 Several other observations are consistent 1950–2002 2009 with the ideas that the availability of chemical probes for a given receptor dictates the 12 level of research interest in it, and that the development of these tools is not driven by the importance of the protein. For instance, 9 large and sustained increases in the rate of publications mentioning a nuclear receptor

6 usually followed, not preceded, the release of a chemical probe. Our findings should serve as a wake-up call to the biomedical and pharmaceutical

Citations (thousands, normalized) 3 research communities. Granting systems must be more daring, institutions must fos- 0 ter and reward risk, and the entire biomedi- Human protein kinases (ordered by the number of publications) cal community must play down the legacy of the literature and let new evidence guide

164 | NaTuRe | VOL 470 | 10 FebRuaRy 2011 © 2011 Macmillan Publishers Limited. All rights reserved COMMENT research. Genome-wide tools such as the University of Toronto, Toronto, Ontario DNA microarrays used in association stud- TOOLS ARE TELLING M5G 1L7, Canada. Ruth Isserlin and ies have allowed geneticists to ignore precon- The availability of research tools in uences a Gary D. Bader are at the Donnelly Centre, ceived ideas about disease mechanisms and protein’s popularity. University of Toronto, Toronto, Ontario pursue a remarkably successful broad-brush 3 M5S 3E1, Canada. Stephen V. Frye is at Proteins for which approach; this approach should be embraced chemical probes the Center for Integrative Chemical Biology more generally. are available and Drug Discovery, Eshelman School of Our data also indicate that high-quality, Proteins for which Pharmacy, University of North Carolina at 2 no probes are readily available research tools can dra- available Chapel Hill, Chapel Hill, North Carolina matically facilitate exploratory biomedical 27599, USA. Timothy M. Willson is at research. Funders such as the Wellcome GlaxoSmithKline, Research Triangle Park,

Trust and the US National Institutes of 1 North Carolina 27709, USA. Frank H. Yu is Health have allocated some funding to tool- in the Program in Neurobiology and Dental Citations (thousands) generating projects, but perhaps not enough. Research Institute, School of Dentistry, Part of the problem is that, unlike the high- Seoul National University, Seoul, 110-749, energy physics community, which endorses 0 South Korea. the creation of large resources, the biomedi- Nuclear hormone receptor e-mail: [email protected] cal community often views projects focused The authors declare competing financial interests: on tool creation with some disdain, for details accompany this article online at go.nature. lacking the elegance of ‘real’ science. and management, as well as collaboration com/kgw3oz. The budgets required also incite a visceral between disciplines, between public and 1. Isserlin, r. et al. Preprint at http://arxiv.org/ reaction. For example, the level of funding private sectors and — to avoid duplication abs/1102.0448 (2011). 2. Grueneberg, D. a. et al. Proc. Natl Acad. Sci. USA needed to develop even one chemical probe of effort — even between countries. 105, 16472–16477 (2008). is enormous. Although it is only a fraction Much of the work that has emerged from 3. Fedorov, o., Müller, s. & Knapp, s. Nature Chem. of the US$100 billion spent on biomedical exploring the human genome over the past Biol. 6, 166–169 (2010). 4. Frye, s. V. Nature Chem. Biol. 6, 159–161 (2010). research each year — about several mil- ten years lies fallow. Challenges notwith- 5. Willson, t. M. & Moore, J. t. Mol. Endocrinol. 16, lion dollars — it is huge compared with the standing, making protein-based research 1135–1144 (2002). amount customarily allocated to an individ- tools readily available must be a major 6. Muscatelli, F. et al. Nature 372, 672–676 (1994). ■ 7. yamagata, K. et al. Nature 384, 458–460 (1996). ual scientist. Finally, the risk is significant. objective in the decade to come. 8. achermann, J. C., Ito, M., Ito, M., hindmarsh, P. C. & Large-scale efforts are not guaranteed to Jameson, J. l. Nature Genet. 22, 125–126 (1999). succeed; they require expertise in science Aled M. Edwards is Banbury professor at the 9. lee, s. l. et al. Science. 273, 1219–1221 (1996).