commentary

Challenges for the ‘chemical-systems’ biologist

Gabriel M Simon & Benjamin F Cravatt

As the field of chemical biology matures, its practitioners are tackling ever more sophisticated biological problems. Chemical approaches, both synthetic and analytical, provide researchers with powerful new technologies to perturb, dissect and even reconstruct complex biological systems. Here we discuss the special challenges and opportunities confronted at the burgeoning interface of chemical and systems biology.

What’s in a name? Depending on one’s per- macromolecules. The ‘new’ distinguish- completed weekly, providing an ever-growing spective, perhaps everything or nothing. By ing characteristic of chemical biology might database on which many large-scale experi- naming or categorizing fields of science, we instead be viewed as one of methodology. A mental approaches, including DNA microar- achieve certain desirable objectives (for exam- conceptually similar distinction can be illus- rays and mass spectrometry–based proteomics, ple, uniting groups of scientists with common trated by reflecting on the historical emergence have been founded. It is becoming increasingly research interests), but these benefits come at of the field of molecular biology. Molecular clear that organizing and interpreting the enor- the cost of creating potentially artificial chasms biology is defined by a set of technologies such mous volumes of data generated by genome- between related disciplines. From journal to as gene amplification, subcloning and sequenc- scale science is a challenge even greater than the departmental titles, simple categorizations ing that, more than any particular research sequencing itself. can result in groups of scientists with com- objective, distinguishes the molecular biolo- Viewing chemical and systems biology mon interests reading distinct sets of litera- gist from other types of scientist. Similarly, by as methodological rather than conceptual ture or being divided into separate buildings, developing and implementing new analytical departures from more traditional disciplines substantially raising the activation barrier for and synthetic methods, chemical biologists underscores the technical prowess and versatil- Nature Publishing Group http://www.nature.com/naturechemicalbiology

8 conversation and collaboration. One might have dramatically expanded the scale and scope ity of these emerging fields and their potential argue that specialization and scientific divi- of biological problems that can be tackled with for cross-fertilization. It is thus perhaps not sion is an inevitable product of our increasing chemical techniques. surprising that new departments have formed © 200 sophistication as researchers. However, the past Similar questions could be asked of systems with the expressed goal of fostering interactions decade has witnessed a growing movement to biology. How does this field, for instance, dif- between chemical and systems biology (for ‘de-differentiate’ research in the life sciences, fer from physiology? We would argue here, example, Stanford University’s Department of engendering “hybrid” disciplines that reflect a too, that the distinction is less about scientific Chemical and Systems Biology and The Scripps desire from within the scientific community objectives, where both fields aim to generate a Research Institute’s Department of Chemical for a resurgence of integrative approaches. holistic understanding of how molecules and Physiology). As cohorts of scientists are built Chief among these emerging (or reemerging) pathways interact to form complex life pro- with a commitment to working at this inter- cross-disciplinary endeavors are chemical biol- cesses, than the approaches that are taken in face, a key question surfaces: what important ogy and systems biology. pursuit of these goals. Systems biology is a dis- biological problems are these bands of inter- Are there features that distinguish chemi- cipline that has emerged from the concordant disciplinary researchers most uniquely suited cal biology from the more classical fields of development of large-scale profiling technolo- to solve? Here, we offer possible answers to biochemistry and pharmacology? These dis- gies and sophisticated computational tools to this question by highlighting several research ciplines certainly share common scientific analyze huge amounts of data. The discovery areas where advances in our understanding of goals—namely to understand and exert chemi- of the genetic code was singularly transfor- ‘systems-level’ biological problems have been cal control over the function of biological mative because it made clear that biology is achieved using chemical approaches. essentially an information science1. It is no Gabriel M. Simon & Benjamin F. Cravatt are at coincidence that the development of power- Chemical approaches for the global The Skaggs Institute for Chemical Biology and ful computers and efficient software went analysis of PTMs the Department of Chemical Physiology, The hand-in-hand with cloning and sequencing Genome sequencing projects have revealed Scripps Research Institute, 10550 North Torrey technologies, culminating in the unraveling that mammals contain surprisingly few genes Pines Road, La Jolla, California 92037, USA. of the human genome in 2001. Now, less than (~20,000), on par with the number found e-mail: [email protected] a decade later, genomes of new organisms are in the comparatively simpler round worm.

nature chemical biology volume 4 number 11 november 2008 639 commentary

Concomitant with this discovery was the real- that exploit the awesome power of mass spec- analytical chemistry platforms is a major driv- ization that gene products (proteins) are found trometry (MS) were used to fully inventory the ing force for much of our modern understand- in a dizzying variety of modified forms in vivo, captured proteins. These approaches further ing of PTMs and their role in biology. leading some to speculate that the human offer a way to assign endogenous substrates to proteome may be comprised of in excess of individual members of huge enzyme classes, Chemical approaches for the global 1,000,000 distinct proteins. Post-translational such as the kinases—a problem that has been analysis of protein activity modifications (PTMs) alter the chemical state challenging to address with genetic approaches Chemical probes that contain a reactive group of proteins in often subtle ways that are not owing to overlapping and/or compensatory that covalently modifies the active sites of easily detected by standard gene or protein enzyme activities. enzymes have engendered the field of activity- profiling techniques. Furthermore, PTMs are Despite these successes, we still lack gen- based protein profiling (ABPP), in which dif- usually dynamic, with the enzymes that add eral profiling methods for many PTMs, not to ferences in protein activity, rather than abun- and remove them having many protein sub- mention techniques to discover new PTMs and dance, can be measured6. The interplay of strates in cells and tissues. This combination connect these events to their cognate enzymes. ABPP with systems biology was founded on of factors highlights the remarkable chemical Will the integrated application of synthetic the premise that probes targeting a broad swath complexity of mammalian proteomes and sug- chemistry, protein engineering and advanced of enzymes could be created by exploiting gests that a comprehensive understanding of MS-based analytical methods prove to be a common mechanistic and/or other active site any PTM will require the ability to perform general strategy for mapping the diversity of features. The ability to profile many enzyme dynamic, systems-level analyses of cells or PTM systems operational in eukaryotic cells? activities in parallel, first by gel6,7 and later organisms. Here, proteomic researchers have We suspect that this will be the case, although by MS6,8, has provided a global view of the enjoyed much success by creating large-scale each PTM offers its own unique challenges. functional state of the proteome in a variety of profiling technologies founded on the prin- Take for instance lysine acetylation and lysine/ physiological and pathological settings. ciples of chemistry. arginine methylation. The ability to identify Early studies with activity-based probes A central challenge for the global analysis of these modifications by MS is not prohibitively taught us that valuable biological information PTMs is the substoichiometric level at which the difficult; what is lacking is a selective capture can be garnered with promiscuous chemical modifications are often found, which is further or enrichment strategy for these modifications. reagents that target many enzymes from a compounded by their chemical and enzymatic It has proven difficult to develop selective anti- given class. These probes were founded on a lability. Key to identifying low-abundance, bodies against acetyl- or methyllysines. Could rich history of knowledge on affinity labels for modified forms of proteins from within the sea researchers engineer acetyltransferases or specific enzyme classes. This concept has been of unmodified proteins has been their covalent methyltransferases that accept modified acetyl- expanded by our lab with the development of capture and enrichment. Chemical biologists CoA or AdoMet groups to permit selective cap- general electrophile probes that label enzymes have pioneered methods to achieve this goal ture and identification of the entire substrate from several mechanistically distinct classes9. for multiple PTMs, including phosphorylation repertoire of individual transferases? Indeed, These electrophilic probes can be thought of and glycosylation. By engineering kinases and certain methyltransferases have been shown as minimalist ‘reactivity probes’ because their glycosyltransferases that are capable of trans- to accept alkyne derivatives of AdoMet4, which labeling is contingent largely on the presence of ferring chemical affinity tags onto phospho- opens up the exciting possibility of transferring activated nucleophiles in the proteome, rather rylated and glycosylated proteins, respectively, alkyne-tagged groups to lysines or arginines than a specific binding pocket or catalytic

Nature Publishing Group http://www.nature.com/naturechemicalbiology 2 3 5

8 the Shokat and Hsieh-Wilson groups have for capture via click chemistry as a strategy mechanism. Initial reports on these reactivity provided general strategies to label, enrich and for substrate identification. When framed in probes, which bear mild carbon electrophiles profile the phospho- and O-GlcNAc proteomes this context, it becomes clear that the interplay such as Michael acceptors, chloroacetamides © 200 with unprecedented breadth and depth (Fig. among bio-orthogonal chemical capture tech- and sulfonate esters, have focused on obtain- 1). In both cases, advanced analytical methods nologies, protein engineering and high-capacity ing inventories of their protein targets and

Enzyme

Tagged modifying Capture chemistry Enzyme group

Engineered enzyme Tagged proteome Post-translationallyost-translati modified proteins

Figure 1 Chemical strategies to profile post-translationally modified proteins with engineered enzymes. An enzyme is engineered to accept and transfer a tagged group (for example, a phosphate or sugar) to its natural substrate within a complex proteome. Subsequently, the tagged protein targets can be covalently captured and enriched for identification. This approach has been used by the Shokat2 and Hsieh-Wilson3 groups to profile the phosphorylated and O-GlcNAc–modified proteomes, respectively.

640 volume 4 number 11 november 2008 nature chemical biology commentary

RG

Target Tagged, irreversible N LC-MS/MS 3 Off-target probe N target profile N N Off-target Capture Off-target chemistry Off-target Off-target Off-target Off-target Complex proteome Reversible inhibitor Selective inhibitor Semiselective inhibitor Nonselective inhibitor

In vivo models

Figure 2 Chemoproteomic strategy to characterize the selectivity of bioactive small molecules in vivo. Covalent inhibitors25 (or reversibly binding small molecules12 that have been derivatized with a reactive group, RG) are converted into activity-based probes via addition of a bio-orthogonal ‘handle’. These probes are added to living systems (cells or animals) and given time to react with protein targets. Probe-labeled proteins are then captured and identified from proteomes using bio-orthogonal chemistry and LC-MS–based proteomic methods, respectively. Using this approach, selective small molecules (red) can be distinguished from compounds with variable degrees of off-target reactivity (green, purple).

specific sites of modification in proteomes, mals) across a broad dose range to fully inven- new features of biological systems? An excel- but the potential for greater systems-level tory their on- and off-targets (Fig. 2). Indeed, lent recent commentary has addressed precisely application is clear. A logical extension of this one might even conclude that it is now more this issue19 and raises the point that protein work would be to attach such tempered car- straightforward to define the proteome-wide depletion, in contrast to chemical inhibition, bon electrophiles, or other tunable reactive specificity of irreversible inhibitors than revers- can lead to discordant phenotypes when pro- elements10,11, to reversibly binding molecules ibly acting compounds. teins perform multiple functions (for instance, to impart upon them the ability to covalently when a protein has both catalytic and scaffold- modify their protein targets (Fig. 2). Successful Chemical genetics: pharmacology meets ing activities that are simultaneously disrupted examples of such a strategy have already systems biology by depletion). Here, we would like to highlight begun to emerge in the literature, resulting for In the 1970s and 1980s, a tremendous amount another often-overlooked special feature of bio- Nature Publishing Group http://www.nature.com/naturechemicalbiology

8 instance in the creation of irreversible inhibi- of biological discovery was driven by forward active small molecules: their ability to perturb tors of Rsk kinases12. genetic screening. Yeast were ideal for these multiple targets in a cellular system. Projecting forward, incorporation of reac- studies owing to their facile genetics and clonal Inhibitors with high target selectivity are © 200 tive elements into bioactive compounds of growth. Forward genetics in higher eukaryotes typically thought of as superior to those that ill-defined mechanism could facilitate charac- is also possible, but is more difficult owing to perturb multiple proteins, and for good rea- terization of their protein targets in proteomes longer lifespans, diploid genomes and difficulty son: off-target effects of pharmaceuticals are or even living systems13,14. These studies might isolating clones. Forward chemical genetics often responsible for toxicity and unexpected in turn reveal that well-designed covalent has, to some degree, picked up where tradi- or misleading biological results. That said, inhibitors display a surprisingly high level of tional forward genetics left off, using a similar sometimes it is precisely the plurality of targets selectivity in the proteome12,15. In this regard, theoretical framework, but employing small- that is responsible for the efficacy of an inhibi- it is interesting to note that the pharmaceuti- molecule libraries to interrogate phenotype tor, and, in this respect, small molecules can cal industry has historically resisted the idea and protein-capture technology to identify reveal the interplay of biochemical pathways of purposefully designing irreversible enzyme targets17. The peculiar challenges associated in complex biological systems that would oth- inhibitors as drugs, owing to worries that they with high-throughput screening of small mol- erwise be missed with selective agents such as might covalently modify other proteins to ecules, not the least of which is target identifica- RNAi. A recent and compelling example is the catastrophic effect16. We would posit, how- tion18, have prompted many researchers to turn drug imatinib (Gleevec), which is used to treat ever, that modern chemical and systems biol- to screening efforts that use molecular biology chronic myelogenous leukemia (CML). The ogy has provided all of the methods required techniques such as RNA interference (RNAi) intended target of imatinib is the fusion pro- to experimentally address this concern in rel- or complementary DNA overexpression. These tein BCR-ABL, which is a constitutively active evant model systems. Covalent inhibitors can methods are certainly valuable and work well to tyrosine kinase that drives proliferation of CML be readily modified with clickable tags that complement and substantiate targets of small cells. Despite its relatively high selectivity for are so sterically small as to make the resulting molecules. However, they also raise a more gen- BCR-ABL over other protein kinases, imatinib activity probes almost indistinguishable from eral question: if phenotypes can be discovered also inhibits other tyrosine kinases, including the original agent. These probes can then be by direct manipulation of mRNA, what value c-Kit. An analog-sensitive (as)-BCR-ABL was applied to any living system (cells or whole ani- does chemical screening hold for discovering engineered that allowed selective inhibition of

nature chemical biology volume 4 number 11 november 2008 641 commentary

this kinase without concomitant inhibition of Looking forward, a number of challenging 1. Crick, F.H., Barnett, L., Brenner, S. & Watts-Tobin, R.J. c-Kit and was used to show that simultaneous problems remain for the ambitious ‘chemical- Nature 192, 1227–1232 (1961). 2. Blethrow, J.D., Glavy, J.S., Morgan, D.O. & Shokat, K.M. inhibition of both kinases was, indeed, required systems’ biologist. Innovative informatic Proc. Natl. Acad. Sci. USA 105, 1442–1447 (2008). for the potent cytotoxic effects of imatinib on strategies are required to integrate and inter- 3. Khidekel, N., Ficarro, S.B., Peters, E.C. & Hsieh-Wilson, leukemia cells20. Thus, the clinical efficacy of pret the huge volumes of data produced by L.C. Proc. Natl. Acad. Sci. USA 101, 13132–13137 (2004). imatinib, lauded as a paragon of rational drug large-scale experimental approaches so that 4. Dalhoff, C., Lukinavicius, G., Klimasauskas, S. & design, is likely due to unintentional blockade coherent descriptions of complex biochemi- Weinhold, E. Nat. Chem. Biol. 2, 31–32 (2006). of multiple targets—a finding that was not fully cal processes can emerge. A glimpse of what 5. Kolb, H.C., Finn, M.G. & Sharpless, K.B. Angew. Chem. Int. Ed. 40, 2004–2021 (2001). appreciated until researchers studied this drug the future may hold can be found in recent 6. Cravatt, B.F., Wright, A.T. & Kozarich, J.W. Annu. Rev. with integrated chemical and systems biology studies where advanced protein engineering22 Biochem. 77, 383–414 (2008). methods. This study and others21 demonstrate and analytical/bioinformatic23 platforms were 7. Greenbaum, D. et al. Mol. Cell. Proteomics 1, 60–68 (2002). that small molecules are often uniquely suited used to generate complementary proteome- 8. Jessani, N. et al. Nat. Methods 2, 691–697 (2005). to drive discovery in complex biological systems wide profiles of proteolytic events in apoptotic 9. Weerapana, E., Simon, G.M. & Cravatt, B.F. Nat. Chem. that may be too robust to be perturbed by single cells. The unification of such experimental Biol. 4, 405–407 (2008). 10. Lee, J., Yu, P., Xiao, X. & Kodadek, T. Mol. Biosyst. 4, target–specific approaches. and computational platforms promises to 59–65 (2008). provide the first holistic view of proteolysis. 11. Salisbury, C.M. & Cravatt, B.F. Proc. Natl. Acad. Sci. USA 104, 1171–1176 (2007). Summary and outlook On the chemical side, facile approaches are 12. Cohen, M.S., Hadjivassiliou, H. & Taunton, J. Nat. Chem. The fields of chemical and systems biology are still needed to generate probes to profile and Biol. 3, 156–160 (2007). poised to tackle complex biological problems perturb biomolecular phenomena, such as 13. Evans, M.J., Saghatelian, A., Sorensen, E.J. & Cravatt, B.F. Nat. Biotechnol. 23, 1303–1307 (2005). owing to the advent of a number of enabling protein-protein interactions, that have histori- 14. Arastu-Kapur, S. et al. Nat. Chem. Biol. 4, 203–213 ‘bridge’ technologies. Chemical capture tech- cally been viewed as “undruggable,” although (2008). niques to enrich specific classes of proteins some headway in this area is being made24. 15. Ahn, K. et al. Biochemistry 46, 13019–13030 (2007). have provided the first global portraits of pro- Ultimately, scientists working at the interface 16. Robertson, J.G. Biochemistry 44, 5561–5571 (2005). tein PTMs and activity in native biological sys- of chemical and systems biology should aspire 17. Knight, Z.A. & Shokat, K.M. Cell 128, 425–430 tems. The utility of chemical probes has been to create a complete tool kit to characterize (2007). 18. Burdine, L. & Kodadek, T. Chem. Biol. 11, 593–597 augmented by advances in bio-orthogonal and and manipulate the functional state of any (2004). analytical chemistry that improve method- protein (or set of related proteins) in native 19. Weiss, W.A., Taylor, S.S. & Shokat, K.M. Nat. Chem. Biol. ological versatility, resolution and sensitivity. biological systems. The proteome-wide imple- 3, 739–744 (2007). 20. Wong, S. et al. Proc. Natl. Acad. Sci. USA 101, 17456– Furthermore, incorporating reactive tags into mentation of these methods has the potential 17461 (2004). small molecules to comprehensively identify to provide the fundamental molecular infor- 21. Chen, S. et al. Proc. Natl. Acad. Sci. USA 103, 17266– their interacting proteins offers a potentially mation required to construct the biochemical 17271 (2006). 22. Mahrus, S. et al. Cell 134, 866–876 (2008). universal strategy to determine mechanism networks of life. As an added bonus, the very 23. Dix, M.M., Simon, G.M. & Cravatt, B.F. Cell 134, 679– of action directly in living systems and might chemical tools that are used to interrogate life 691 (2008). one day even make a case for the preferential systems should also facilitate drug develop- 24. Wells, J.A. & McClendon, C.L. Nature 450, 1001–1009 (2007). adoption of covalent agents as pharmacologi- ment efforts to treat these systems when they 25. Alexander, J.P. & Cravatt, B.F. Chem. Biol. 12, 1179– cal probes and drugs. go awry. 1187 (2005). Nature Publishing Group http://www.nature.com/naturechemicalbiology 8 © 200

642 volume 4 number 11 november 2008 nature chemical biology