- Email: [email protected]
Selecting chemicals: the emerging utility of DNA-encoded libraries
Available online at www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Matthew A Clark Over the past 10 years, a handful of academic and industrial research groups have developed strategies for the synthesis and interrogation of DNA-encoded small-molecule libraries. These strategies can be divided into those in which DNA directs small-molecule synthesis and those in which it records the synthesis. These libraries have started to yield novel modulators of biological targets, including: SH3-domainbinding peptoids, macrocyclic peptide-based Bcl-XL/BH3 interaction disruptors, ligands for TNF, albumin, streptavidin and others, and small-molecule kinase inhibitors. The DNAencoded library field holds the potential to address the general problem of biological ligand discovery, including pharmaceutical lead generation. Address GlaxoSmithKline, Molecular Discovery Research, 830 Winter St. Waltham, MA 02451, United States Corresponding author: Clark, Matthew A. ([email protected])
Current Opinion in Chemical Biology 2010, 14:396–403 This review comes from a themed issue on Molecular Diversity Edited by Lisa A. Marcaurelle and Mike Foley Available online 26th March 2010 1367-5931/$ – see front matter # 2010 Elsevier Ltd. All rights reserved. DOI 10.1016/j.cbpa.2010.02.017
Introduction The discovery of functional ligands to biological protein targets is fundamental to understanding living systems and discovering new therapeutics. While numerous methods are available in this search, they can be roughly divided into two categories: chemical screening and biochemical selection. Screening is the repetitive interrogation of individual compounds for a desired functional property. Compound screening has been performed since humans first explored the benefits (and dangers) of natural extracts. Later, when chemists became able to create novel compounds, it became routine to test new substances for various biological properties. In its largescale application, screening is the institutionalization of serendipity, an attempt to repeat the unexpected discoveries of penicillin or cisplatin in a controlled environment. Since the likelihood is small that any particular compound will possess a desired activity, large numbers of comCurrent Opinion in Chemical Biology 2010, 14:396–403
pounds are required to achieve a viable success rate. In the pharmaceutical industry, such high-throughput screening (HTS) is the foundational hit-generating technology. Corporate collections containing over 1 million compounds are routinely screened. A biological approach to discovery is different, in that it harnesses Darwinian selection to discover the desired entity. Very large populations of candidates are subjected to selective pressure, allowing only the fittest members to survive. Selection was first applied to organisms, from the prehistoric selection of the most bountiful strains of wild wheat to the isolation of transfected E. coli cells by antibiotic resistance. The advent of molecular biology, however, allowed selection to be conducted at the molecular level. Techniques such as phage display, aptamer SELEX, and mRNA display and ribosome display enable the in vitro evolution of biological macromolecules. These evolutionary processes are truly Darwinian in that they allow mutation during amplification steps, further diversifying the selectable populations. Biomolecular evolution techniques rely on the intimate connection of phenotype and genotype, either by encapsulation (phage), covalent attachment (mRNA display), or by combining the genotypic and phenotypic functions into a single substance, nucleic acid (aptamers and ribozymes). Selection offers immense advantages over screening in terms of numbers, flexibility, convenience, and cost. But selection traditionally suffered a fatal shortcoming, at least for pharmaceutical discovery: it could only be applied to macromolecules of the biological central dogma: DNA, RNA, and proteins. Selection of unnatural molecules requires methods for associating a synthetic chemotype with a readable genotype. DNA, with its stability and amenablility to manipulation, amplification, and sequencing, is the obvious choice for the genotypic element. This fact was recognized by Lerner and Brenner [1], and independently by researchers at Affymax [2], who first described a strategy for DNA encoded synthesis. A variety of methods for associating DNA with a chemotype have been developed over the past ten years. This review will survey current approaches to DNA-encoded chemistry, with a focus on combinatorial synthesis of DNA-encoded libraries (DELs) and the selection of such libraries for pharmaceutically relevant compounds.
Sequence-programmed DELs A new approach to DNA-encoded chemistry was reported in the pioneering work of Gartner and Liu in 2001 [3]. www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 397
Figure 1
Preparation of DELs by DNA-programmed chemistry. (a) DNA-templated synthesis: annealing of DNA brings together oligo-linked reactants. Selection output can be amplified by PCR, to serve as template for synthesis of an enriched sublibrary. Reproduced with permission of Prof. D. Liu. (b) The Vipergen yoctoreactor: chemistry is confined in a three-way DNA junction. Amplification of selection output by ‘rolling amplification’ is not shown [10]. Reproduced with permission, #2009, American Chemical Society. (c) DNA-routed synthesis: sequence complementarity localizes starting materials to individual reaction vessels for split-and-pool synthesis. As with templated synthesis, selection output can be amplified by PCR and library resynthesis [14]. Reproduced with permission, #2007, American Chemical Society.
The Liu group developed DNA-templated synthesis, in which sequence complementarity brings into proximity DNA-appended reactants [4]. Pseudo-intramolecularity ensures acceptable rates and yields of the bond-forming steps (see Figure 1a). Bond formation is followed by removal of the DNA strand from one of the reactants, regenerating ssDNA for entry into the next synthetic step. The strategy allows for one-pot library synthesis [5] and, in principal, amplification of library members by PCR and resynthesis. A variety of chemical transformations have been developed [6] and multi-step synthesis conducted to give libraries of moderate size [7]. Liu and co-workers were the first to demonstrate affinity selection of DNA–small-molecule conjugates, through ‘spike-in’ www.sciencedirect.com
experiments using known ligands linked to DNA [8]. The selection method was similar to earlier SELEX work, involving immobilization of the protein target on polymer matrix and affinity-based partitioning of the library. Despite its seminal contributions to the field, the Liu group has not reported the discovery of novel ligands from DEL selection. Ensemble Discovery (Cambridge, MA), which has licensed the Liu method, has reported the discovery of Bcl-XL/BH3 interaction disruptors from a DNA-encoded library of peptide macrocycles (http:// www.ensemblediscovery.com/news/index.html). A further extension of DNA-templated chemistry was described by workers at Vipergen ApS (Copenhagen, Current Opinion in Chemical Biology 2010, 14:396–403
398 Molecular Diversity
Denmark) and collaborators [9]. This work involves the formation of three-dimensional DNA junctions to bring the reactants into close proximity in a confined environment (Figure 1b). Similar to the work of Liu, the Vipergen team was able to demonstrate enrichment of a known ligand from a synthesized library. Notably, they demonstrated amplification of selection output through an ingenious process they named ‘rolling translation.’ This work provides a glimpse of what is possible when one judiciously exploits the flexible molecular architecture of DNA. One challenge of templated DEL chemistry is the need to attach building blocks to DNA before library synthesis. In a series of papers published in 2004, Halpin and Harbury reported a solution to this problem (Figure 1c) [10]. They developed a method of DNA-controlled routing of starting materials to separate reaction vessels [11]. Complementarity to resin-bound ssDNA separates DNAtagged templates into various compartments where solidphase synthesis can be performed [12]. Using traditional split-and-pool strategy, the Harbury method allows one reaction per vessel, rather than all reactions in one pot as with templated synthesis. This means that off-the-shelf reagents may be utilized. In a landmark study [13], Harbury and co-workers used this method to construct a 100-million-compound library of 8-mer peptoids, the largest DNA-encoded synthetic library at that time. The library was successfully interrogated over 6 rounds of selection to discover previously unknown peptoid ligands for an SH3 domain. In contrast to the targetimmobilized selection strategy described above, the SH3 selection employed incubation of library with soluble target, followed by antibody-based capture. Several
related peptoids were discovered, with affinities ranging from 16 to 98 mM. A notable feature of this work was the iterative amplification of the library following selection steps.
Sequence-recorded DELs The three techniques cited above are similar in that they use DNA to control the synthetic destiny of library members. Analogous to biological transcription and translation, the encoding DNA sequences are assembled before library synthesis, and subsequently used to program molecular synthesis. An alternative strategy is to build the DNA code during the course of library synthesis. Rather than their destiny, the DNA holds the synthetic history of the library members. Concomitant chemical synthesis of encoding DNA and library molecules was the basis of the original proposal of Lerner and Brenner. Kinoshita and Nishigaki later suggested enzymatic assembly of encoding DNA as an alternative to chemical synthesis [14]. The work discussed below exclusively employs enzymatic DNA assembly during library synthesis. The group of Neri and co-workers has developed a diverse research program in DNA-encoded discovery methods, including DEL synthesis. They have reported the synthesis of several DELs via multi-step schemes, both of which employ two cycles of diversity generation, giving total library sizes of 4000 compounds [15,16]. The encoding strategy involves Klenow-catalyzed DNA polymerization, templated by partially complementary coding strands (Figure 2a). Amide bond formation and Diels–Alder cycloaddition were used to generate DEL diversity.
Figure 2
Preparation of DELs by DNA-recorded synthesis. (a) Synthesis of 4000 Diels–Alder products with encoding by DNA polymerization. The initial DNA– small-molecule conjugate is single stranded. An accessory strand serves as template for Klenow-based synthesis of the cycle 2 code. The final product is double stranded [16,19]. Reproduced with permission, #2008 and 2009, Elsevier, Ltd. (b) Synthesis of 800 million triazine amides using ligation-based encoding. The starting material is covalently linked double-stranded DNA. Each synthetic step is accompanied by enzymatic installation of double-stranded tags [21]. Current Opinion in Chemical Biology 2010, 14:396–403
www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 399
Libraries of this type cannot be amplified and subjected to many rounds of iterative selection. The selection process therefore must be efficient and the sampling of the output population thorough. The Neri group recognized that modern high-capacity sequencing techniques, such as that of 454 Technologies [17], could be a powerful tool for the characterization of selection output [16]. The low enrichments achieved after only 1–3 rounds of selections are only observable if tens of thousands of sequences are obtained (Figure 3a). Especially with increasing library sizes (see GSK work below), one can appreciate that characterization of selection output from non-amplifiable libraries would have been impossible with the cloning and Sanger sequencing methods available only 5 years ago. In this context DEL technology can be seen as the chemical application of cutting-edge advancements in molecular biology and genomic sciences. With over 40,000 sequences obtained per experiment, the Neri group was able to identify binders for streptavidin, polyclonal IgG, MMP-3, TNF, Bcl-xL, and albumin after a single round of selection [16,18]. One of the TNF ligands was reported to be active in a TNF cytotoxicity assay. The company Philochem (Zurich, Switzerland) has filed a patent application on the Neri methodology and appears to be utilizing the method for pharmaceutical hit identification (http://www.philochem.ch). We at the former Praecis Pharmaceuticals (Waltham, MA), now part of GlaxoSmithKline, have operated DNA-encoded chemistry for lead discovery since 2003. In a recent report, we disclosed a method of encoding that utilized covalently linked double-stranded DNA, in which dsDNA tags were installed during library synthesis by enzymatic ligation [19]. A DEL of 800 million compounds was constructed by 4 cycle split-and-pool synthesis (Figure 2b), and the 3-round selection output was sequenced by 454 sequencing (Figure 3c). Novel inhibitors of p38 MAP kinase with low nanomolar activity were discovered. Crystallography of enzyme-inhibitor complexes revealed how the tethered DNA could be tolerated in the binding interaction. Our publication was the first to report discovery of novel and potent small-molecule enzyme inhibitors from a DNA encoded library. The library was orders of magnitude larger than most other DNA-encoded libraries reported to date, with the exception of the Harbury peptoid library. The GSK report is also notable in that it describes how highthroughout sampling of the selection output revealed structural trends among the selected compounds, allowing identification of the pharmacophoric substructure. GSK continues to utilize the technology for its internal lead discovery programs. Nuevolution AB (Copenhagen, Denmark) has been practicing DNA-encoded technologies since 2001. The comwww.sciencedirect.com
pany has reported both a DNA-templating approach similar to Liu and Ensemble, and enzymatic construction of encoding DNA similar to Neri and GSK. Nuevolution has announced collaborations with Merck, Novartis, and Lexicon Pharmaceuticals (http://www.nuevolution.com/ News/Press_releases.htm).
Prospects The prospects of DEL technology for pharmaceutical lead generation appear promising. The literature reviewed herein show that DEL mechanics function as designed and allow rapid and convenient ligand discovery. Whether this facility can make an impact upon pharmaceutical hit discovery and progression to drug candidates remains to be seen. With the industry increasingly concerned with physical properties and their relationship to candidate attrition, DEL outputs will have to conform to standards of lead-likeness and tractability. Current examples of compounds discovered by DEL selection, comprising >600 MW heterocycles, peptoids, cyclic peptides, and large, lipophilic, flexible structures (see Figure 4), illustrate that improvement is needed. Fortunately, the physico-chemical properties of DELs are under the control of the library chemist, through the synthetic schemes and building blocks employed. As a further challenge, the nature of affinity selection precludes a variety of powerful screening paradigms, such as functional, phenotypic, and black-box screening. It is likely that compounds found simply on the basis of binding will frequently lack a desired function. It therefore is crucial that selection methods and strategies be developed to weed out these non-functional binders early in the process. Lastly, it is obvious that DNA-tethered molecules have fewer possible binding modes than free molecules, since the DNA and linker may interfere with target binding. While sufficient library diversity may compensate for this limitation in some cases, it is likely that some targets, such as those with deep and inaccessible binding sites, will be inherently unamenable to a DNA-linked approach. As pointed out in [13], DEL technology allows one to explore the relationship between numeric diversity and biological activity, in chemical space removed from the biopolymer area. Our data at GSK appear to indicate that larger libraries do indeed have higher rates of success. In [19], two triazine libraries were described, DEL-A containing 7 million compounds, and DEL-B with 800 million. In our hands, DEL-B has proven much more productive, furnishing submicromolar inhibitors to multiple kinases, a metalloprotease, a metal-dependent hydroxylase, and several hydrolases including soluble epoxide hydrolase. DEL-A in contrast has given little beyond a few kinase inhibitors. Of course, this may be due to DEL-B having some structural property that makes it more likely to bind biological targets, akin to the ‘privileged scaffolds’ concept in medicinal chemistry Current Opinion in Chemical Biology 2010, 14:396–403
400 Molecular Diversity
Figure 3
Current Opinion in Chemical Biology 2010, 14:396–403
www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 401
Figure 4
Outputs from DEL selection reported in the literature. (a) Peptoids selected for binding to N-CrkSH3. Dissociation constants were determined by tryptophan fluorescence perturbation [14]. (b) Kinase inhibitors discovered by selection against Aurora A and p38 MAP kinases. IC50 values were determined by standard kinase inhibition assay [21]. (c) Ligands for streptavidin, albumin, TNF, and Bcl-xL discovered by selection of two 4000compound DNA-encoded libraries. Dissociation constants were determined by fluorescence polarization assay using fluorescein-labeled library compounds [17,19].
(Legend Figure 3) Methods of visualizing chemical space: characterization of encoded library selection output. (a) Frequency distribution of ca. 40,000 sequences obtained after selection of 4000-compound 2-cycle library against streptavidin. Bar heights correspond to the frequency of a given library member. Inset are plots of the library before selection and after selection against naked resin. Structures with high frequency of occurrence were resynthesized and tested for streptavidin binding [17]. #2008, National Academy of Science, U.S.A. (b) Clustergram of 100-million-compound 8cycle peptoid library selected against the N-CrkSH3 domain. Of 960 clones, 840 full-length reads were obtained; these encompassed 215 unique peptoid sequences, of which 112 occurred more than once. The sequences were clustered by sequence similarity and fell into 10 families [14]. Reproduced with permission, #2007, American Chemical Society. (c) Representation of a family from 800-million-compound 4-cycle library selected against p38 MAP kinase. Since the library had four dimensions of diversity, an array of three-dimensional cube plots was necessary to fully visualize the library space. The cube shown corresponds to the sublibrary containing benzimidazole-5-carboxylic acid at cycle 2. Low-occurring molecules have been removed to reveal an enriched family (red = 2 copies, black = 3 copies, green = 4 copies) [21]. www.sciencedirect.com
Current Opinion in Chemical Biology 2010, 14:396–403
402 Molecular Diversity
[20]. Untangling the interplay of chemical class, numerical diversity, and biological activity will be an exciting theme as the field progresses.
exploration exists at present. One hopes that the growth of the DEL field will inspire research groups to pursue the challenge of on-DNA synthetic chemistry.
Interrogation of libraries through en masse affinity selection has several important advantages over functional screening of discretes. First, the time, cost, and material consumed are greatly reduced. Second, because interrogation is conducted by simple partitioning, spurious results due to aggregation, autofluorescence, and insolubility are minimized. Third, the format enables one to conduct a variety of parallel experiments, such as: competition with known ligands; cross-selection with isoforms or selectivity targets; addition or subtraction of co-factors, substrates, or binding partners; and titration of target concentration. The output of such a campaign can generate hypotheses about selectivity, binding sites, dissociation constants, and biological function before compounds have been synthesized and assayed. We have found this data package empowers the hit-to-lead chemists who prosecute the off-DNA chemistry.
In summary, DNA-encoded chemistry appears to be a growing field, whose promise of fast, cheap, and reliable ligand discovery is starting to be realized. The DEL technology promises a future where a single scientist can interrogate a billion compounds in a few hours, where structure–activity relationships and selectivity are determined in a primary screen, and where the chemical space we choose to explore is determined by design rather than by history. Perhaps, with appropriate library design and cross-selection experiments, advanced leads or even drug candidates could be obtained directly from selection. Only time will tell whether this technology becomes a useful tool of pharmaceutical drug discovery. In the meantime, the dream of evolving synthetic substances as if they were biomolecules has never been closer to reality.
Acknowledgements A final advantage of affinity selection is the SAR data created when large chemical collections are subjected to simultaneous interrogation. Families of compounds are invariably observed; indeed, a lack of familial relationships among the selected compounds could be considered an indication of poor selection. The selection of closely related structures, including stereoisomers and isosteres, provides confidence that a structurally driven binding event is occurring. By examining the commonalities and divergences among the family members, one can surmise what is the key pharmacophore, and what regions could be further optimized or removed. A possible consequence of this large-scale SAR examination is that it makes DEL technology more tolerant of synthetic heterogeneity than traditional methods. Selected compounds that are the products of reaction failure can be identified by certain trends in the selection data, a fortuitous observation since characterization of mixtures of this size remains a challenge. This fact puts DEL technology at odds with traditional chemistry concerns about yield, homogeneity, and characterization. How this balance is addressed over the coming years will be a fascinating aspect of DEL technology. A common critique of DEL technology is that the aqueous and DNA-compatible conditions required for the chemistry prevent the synthesis of diverse and useful structures. However, examination of the survey by Liu [6] shows that most of the useful reactions for traditional bead-based array synthesis are amenable to DNAencoded synthesis: amide formation, reductive amination, Pd-catalyzed cross-coupling, SNAr, dipolar cycloaddition, and protecting group manipulation. This does not mean that the scope of DEL chemistry needs no further expansion, only that a strong foundation for chemical Current Opinion in Chemical Biology 2010, 14:396–403
The author is grateful to the following colleagues for their help in preparing this manuscript: Barry Morgan, Christopher Arico-Muendel, Todd Graybill, Bryan King, Jeffrey Gross, and Jeff Sutton.
References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest of outstanding interest 1. Brenner S, Lerner RA: Encoded combinatorial chemistry. Proc Natl Acad Sci USA 1992, 89:5381-5383. This theoretical analysis of an encoding technology is the first paper to lay out the rationale, the challenges, and the possibilities of a synthetic molecule selection technique. The discussions of tag design, chemical compatibility, selection techniques, PCR, and amplification remain relevant today. 2.
Needels MC, Jones DG, Tate EH, Heinkel GL, Kochersperger LM, Dower WJ, Barrett RW, Gallop MA: Generation and screening of an oligonucleotide-encoded synthetic peptide library. Proc. Natl. Acad. Sci USA 1993, 90:10700-10704.
3.
Gartner ZJ, Liu DR: The generality of DNA-templated synthesis as a basis for evolving non-natural small molecules. J Am Chem Soc 2001, 123:6961-6963. With this paper, Liu and co-workers first illustrate their concept for a DNAtemplated chemistry method. Using simple nucleophilic addition and substitution reactions, the basic principles of sequence-dependent reactivity, hybridization architecture, and distance effects are elucidated, forming a foundation for later work.
4.
Gartner ZJ, Kanan MW, Liu DR: Multistep small-molecule synthesis programmed by DNA templates. J Am Chem Soc 2002, 124:10304-10306. Building on the earlier work, Liu and co-workers show for the first time that multi-step, template-directed synthetic schemes were feasible. The scope of the chemistry and linker strategies employed reveal that rich structural diversity is possible in a DNA-templated context. 5.
Snyder TM, Liu DR: Ordered multistep synthesis in a single solution directed by DNA templates. Angew Chem Int Ed Engl 2005, 44:7379-7382.
6.
Gartner ZJ, Kanan MW, Liu DR: Expanding the reaction scope of DNA-templated synthesis. Angew Chem Int Ed Engl 2002, 41:1796-1800. Putting to rest any concerns about the diversity of the DNA-templated synthetic repertoire, Liu and co-workers report a wide range of feasible www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 403
synthetic transformations, including Pd-catalyzed cross-coupling, dipolar cycloadditions, reductive amination, and Wittig olefination. This report remains a benchmark and a challenge for the generation of DNA-encoded chemical diversity. 7.
Tse BN, Snyder TM, Shen Y, Liu DR: Translation of DNA into a library of 13,000 synthetic small-molecule macrocycles suitable for in vitro selection. J Am Chem Soc 2008, 130:15611-15626.
8.
Doyon JB, Snyder TM, Liu DR: Highly sensitive in vitro selections for DNA-linked synthetic small molecules with protein binding affinity and specificity. J Am Chem Soc 2003, 125:12372-12373. A DNA-encoded library technology is useless without a means of selection; in this paper Liu and co-workers demonstrate for the first time the feasibility of DEL selection. A variety of known ligands are enriched from large populations to varying degrees, over a single selection round.
9.
Hansen MH, Blakskjaer P, Petersen LK, Hansen TH, Hojfeldt JW, Gothelf KV, Hansen NJ: A yoctoliter-scale DNA reactor for small-molecule evolution. J Am Chem Soc 2009, 131:1322-1327. In this paper, a truly original and ingenious manipulation of DNA architecture is employed to generate DNA-templated chemical libraries. Especially notable is the amplification technique, called rolling translation, which was reduced to practice and used to enrich a known ligand over 2 rounds of selection. 10. Halpin DR, Harbury PB: DNA display I. Sequence-encoded routing of DNA populations. PLoS Biol 2004, 2:E173. 11. Halpin DR, Harbury PB: DNA display II. Genetic manipulation of combinatorial chemistry libraries for small-molecule evolution. PLoS Biol 2004, 2:E174. 12. Halpin DR, Lee JA, Wrenn SJ, Harbury PB: DNA display III. Solid-phase organic synthesis on unprotected DNA. PLoS Biol 2004, 2:E175. Building on concepts delineated in the two earlier papers, this report describes a novel strategy for DNA-encoded synthesis based on sequence-controlled routing of library members. The method of Harbury and co-workers is the only DNA-encoding technology that utilizes solidphase synthesis for library generation. 13. Wrenn SJ, Weisinger RM, Halpin DR, Harbury PB: Synthetic ligands discovered by in vitro selection. J Am Chem Soc 2007, 129:13137-13143. In this landmark report, Harbury and co-workers describe the synthesis of a 100-million member peptoid library using their sequence routing technique. They then interrogate the library for SH3-domain binders through 6 rounds of affinity selection, and sequence the output to identify several families of putative binders. Ligands with micromolar affinities are dis-
www.sciencedirect.com
covered. This work represents the first example of novel ligand discovery from a DNA-encoded library. 14. Kinoshita Y, Nishigaki K: Enzymatic synthesis of code regions for encoded combinatorial chemistry (ECC). Nucleic Acids Symp Series 1995, 34:201-202. Recognizing the difficulties of in situ chemical synthesis of oligonucleotides, the authors suggest an enzymatic strategy for construction of encoding DNA, a technique that is now standard in DNA-recorded synthesis efforts. 15. Buller F, Mannocci L, Zhang Y, Dumelin CE, Scheuermann J, Neri D: Design and synthesis of a novel DNA-encoded chemical library using Diels–Alder cycloadditions. Bioorg Med Chem Lett 2008, 18:5926-5931. 16. Mannocci L, Zhang Y, Scheuermann J, Leimbacher M, De Bellis G, Rizzi E, Dumelin C, Melkko S, Neri D: High-throughput sequencing allows the identification of binding molecules isolated from DNA-encoded chemical libraries. Proc Natl Acad Sci USA 2008, 105:17670-17675. Realizing the need for deep sequencing in characterizing selection output, the Neri group employs the 454 Technologies sequencing platform. Thorough sampling of the selection output allows identification of modest ligands to a variety of targets from a 4000-compound library. This report is the second demonstration of novel ligand discovery from a naı¨ve DNAencoded library. 17. Margulies M, Egholm M, Altman WE, Attiya S, Bader JS, Bemben LA, Berka J, Braverman MS, Chen YJ, Chen Z et al.: Genome sequencing in microfabricated high-density picolitre reactors. Nature 2005, 437:376-380. 18. Buller F, Zhang Y, Scheuermann J, Schafer J, Buhlmann P, Neri D: Discovery of TNF inhibitors from a DNA-encoded chemical library based on Diels–Alder cycloaddition. Chem Biol 2009, 16:1075-1086. Building on the previous work, Neri and co-workers report discovery of additional ligands to novel targets using their 4000-compound library based on Diels–Alder chemistry. 19. Clark MA, Acharya RA, Arico-Muendel CC, Belyanskaya SL, Benjamin DR, Carlson NR, Centrella PA, Chiu CH, Creaser SP, Cuozzo JW et al.: Design, synthesis and selection of DNA-encoded small-molecule libraries. Nat Chem Biol 2009, 5:647-654. This report from the GSK group describes the largest DNA-encoded library yet disclosed at 800 million members. Using high-throughput sequencing, kinase inhibitors with nanomolar potency are discovered. This is the first report of novel and potent small-molecule enzyme inhibitors from a DNA-encoded library. 20. Costantino L, Barlocco D: Privileged structures as leads in medicinal chemistry. Curr Med Chem 2006, 13:65-85.
Current Opinion in Chemical Biology 2010, 14:396–403
Selecting chemicals: the emerging utility of DNA-encoded libraries Matthew A Clark Over the past 10 years, a handful of academic and industrial research groups have developed strategies for the synthesis and interrogation of DNA-encoded small-molecule libraries. These strategies can be divided into those in which DNA directs small-molecule synthesis and those in which it records the synthesis. These libraries have started to yield novel modulators of biological targets, including: SH3-domainbinding peptoids, macrocyclic peptide-based Bcl-XL/BH3 interaction disruptors, ligands for TNF, albumin, streptavidin and others, and small-molecule kinase inhibitors. The DNAencoded library field holds the potential to address the general problem of biological ligand discovery, including pharmaceutical lead generation. Address GlaxoSmithKline, Molecular Discovery Research, 830 Winter St. Waltham, MA 02451, United States Corresponding author: Clark, Matthew A. ([email protected])
Current Opinion in Chemical Biology 2010, 14:396–403 This review comes from a themed issue on Molecular Diversity Edited by Lisa A. Marcaurelle and Mike Foley Available online 26th March 2010 1367-5931/$ – see front matter # 2010 Elsevier Ltd. All rights reserved. DOI 10.1016/j.cbpa.2010.02.017
Introduction The discovery of functional ligands to biological protein targets is fundamental to understanding living systems and discovering new therapeutics. While numerous methods are available in this search, they can be roughly divided into two categories: chemical screening and biochemical selection. Screening is the repetitive interrogation of individual compounds for a desired functional property. Compound screening has been performed since humans first explored the benefits (and dangers) of natural extracts. Later, when chemists became able to create novel compounds, it became routine to test new substances for various biological properties. In its largescale application, screening is the institutionalization of serendipity, an attempt to repeat the unexpected discoveries of penicillin or cisplatin in a controlled environment. Since the likelihood is small that any particular compound will possess a desired activity, large numbers of comCurrent Opinion in Chemical Biology 2010, 14:396–403
pounds are required to achieve a viable success rate. In the pharmaceutical industry, such high-throughput screening (HTS) is the foundational hit-generating technology. Corporate collections containing over 1 million compounds are routinely screened. A biological approach to discovery is different, in that it harnesses Darwinian selection to discover the desired entity. Very large populations of candidates are subjected to selective pressure, allowing only the fittest members to survive. Selection was first applied to organisms, from the prehistoric selection of the most bountiful strains of wild wheat to the isolation of transfected E. coli cells by antibiotic resistance. The advent of molecular biology, however, allowed selection to be conducted at the molecular level. Techniques such as phage display, aptamer SELEX, and mRNA display and ribosome display enable the in vitro evolution of biological macromolecules. These evolutionary processes are truly Darwinian in that they allow mutation during amplification steps, further diversifying the selectable populations. Biomolecular evolution techniques rely on the intimate connection of phenotype and genotype, either by encapsulation (phage), covalent attachment (mRNA display), or by combining the genotypic and phenotypic functions into a single substance, nucleic acid (aptamers and ribozymes). Selection offers immense advantages over screening in terms of numbers, flexibility, convenience, and cost. But selection traditionally suffered a fatal shortcoming, at least for pharmaceutical discovery: it could only be applied to macromolecules of the biological central dogma: DNA, RNA, and proteins. Selection of unnatural molecules requires methods for associating a synthetic chemotype with a readable genotype. DNA, with its stability and amenablility to manipulation, amplification, and sequencing, is the obvious choice for the genotypic element. This fact was recognized by Lerner and Brenner [1], and independently by researchers at Affymax [2], who first described a strategy for DNA encoded synthesis. A variety of methods for associating DNA with a chemotype have been developed over the past ten years. This review will survey current approaches to DNA-encoded chemistry, with a focus on combinatorial synthesis of DNA-encoded libraries (DELs) and the selection of such libraries for pharmaceutically relevant compounds.
Sequence-programmed DELs A new approach to DNA-encoded chemistry was reported in the pioneering work of Gartner and Liu in 2001 [3]. www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 397
Figure 1
Preparation of DELs by DNA-programmed chemistry. (a) DNA-templated synthesis: annealing of DNA brings together oligo-linked reactants. Selection output can be amplified by PCR, to serve as template for synthesis of an enriched sublibrary. Reproduced with permission of Prof. D. Liu. (b) The Vipergen yoctoreactor: chemistry is confined in a three-way DNA junction. Amplification of selection output by ‘rolling amplification’ is not shown [10]. Reproduced with permission, #2009, American Chemical Society. (c) DNA-routed synthesis: sequence complementarity localizes starting materials to individual reaction vessels for split-and-pool synthesis. As with templated synthesis, selection output can be amplified by PCR and library resynthesis [14]. Reproduced with permission, #2007, American Chemical Society.
The Liu group developed DNA-templated synthesis, in which sequence complementarity brings into proximity DNA-appended reactants [4]. Pseudo-intramolecularity ensures acceptable rates and yields of the bond-forming steps (see Figure 1a). Bond formation is followed by removal of the DNA strand from one of the reactants, regenerating ssDNA for entry into the next synthetic step. The strategy allows for one-pot library synthesis [5] and, in principal, amplification of library members by PCR and resynthesis. A variety of chemical transformations have been developed [6] and multi-step synthesis conducted to give libraries of moderate size [7]. Liu and co-workers were the first to demonstrate affinity selection of DNA–small-molecule conjugates, through ‘spike-in’ www.sciencedirect.com
experiments using known ligands linked to DNA [8]. The selection method was similar to earlier SELEX work, involving immobilization of the protein target on polymer matrix and affinity-based partitioning of the library. Despite its seminal contributions to the field, the Liu group has not reported the discovery of novel ligands from DEL selection. Ensemble Discovery (Cambridge, MA), which has licensed the Liu method, has reported the discovery of Bcl-XL/BH3 interaction disruptors from a DNA-encoded library of peptide macrocycles (http:// www.ensemblediscovery.com/news/index.html). A further extension of DNA-templated chemistry was described by workers at Vipergen ApS (Copenhagen, Current Opinion in Chemical Biology 2010, 14:396–403
398 Molecular Diversity
Denmark) and collaborators [9]. This work involves the formation of three-dimensional DNA junctions to bring the reactants into close proximity in a confined environment (Figure 1b). Similar to the work of Liu, the Vipergen team was able to demonstrate enrichment of a known ligand from a synthesized library. Notably, they demonstrated amplification of selection output through an ingenious process they named ‘rolling translation.’ This work provides a glimpse of what is possible when one judiciously exploits the flexible molecular architecture of DNA. One challenge of templated DEL chemistry is the need to attach building blocks to DNA before library synthesis. In a series of papers published in 2004, Halpin and Harbury reported a solution to this problem (Figure 1c) [10]. They developed a method of DNA-controlled routing of starting materials to separate reaction vessels [11]. Complementarity to resin-bound ssDNA separates DNAtagged templates into various compartments where solidphase synthesis can be performed [12]. Using traditional split-and-pool strategy, the Harbury method allows one reaction per vessel, rather than all reactions in one pot as with templated synthesis. This means that off-the-shelf reagents may be utilized. In a landmark study [13], Harbury and co-workers used this method to construct a 100-million-compound library of 8-mer peptoids, the largest DNA-encoded synthetic library at that time. The library was successfully interrogated over 6 rounds of selection to discover previously unknown peptoid ligands for an SH3 domain. In contrast to the targetimmobilized selection strategy described above, the SH3 selection employed incubation of library with soluble target, followed by antibody-based capture. Several
related peptoids were discovered, with affinities ranging from 16 to 98 mM. A notable feature of this work was the iterative amplification of the library following selection steps.
Sequence-recorded DELs The three techniques cited above are similar in that they use DNA to control the synthetic destiny of library members. Analogous to biological transcription and translation, the encoding DNA sequences are assembled before library synthesis, and subsequently used to program molecular synthesis. An alternative strategy is to build the DNA code during the course of library synthesis. Rather than their destiny, the DNA holds the synthetic history of the library members. Concomitant chemical synthesis of encoding DNA and library molecules was the basis of the original proposal of Lerner and Brenner. Kinoshita and Nishigaki later suggested enzymatic assembly of encoding DNA as an alternative to chemical synthesis [14]. The work discussed below exclusively employs enzymatic DNA assembly during library synthesis. The group of Neri and co-workers has developed a diverse research program in DNA-encoded discovery methods, including DEL synthesis. They have reported the synthesis of several DELs via multi-step schemes, both of which employ two cycles of diversity generation, giving total library sizes of 4000 compounds [15,16]. The encoding strategy involves Klenow-catalyzed DNA polymerization, templated by partially complementary coding strands (Figure 2a). Amide bond formation and Diels–Alder cycloaddition were used to generate DEL diversity.
Figure 2
Preparation of DELs by DNA-recorded synthesis. (a) Synthesis of 4000 Diels–Alder products with encoding by DNA polymerization. The initial DNA– small-molecule conjugate is single stranded. An accessory strand serves as template for Klenow-based synthesis of the cycle 2 code. The final product is double stranded [16,19]. Reproduced with permission, #2008 and 2009, Elsevier, Ltd. (b) Synthesis of 800 million triazine amides using ligation-based encoding. The starting material is covalently linked double-stranded DNA. Each synthetic step is accompanied by enzymatic installation of double-stranded tags [21]. Current Opinion in Chemical Biology 2010, 14:396–403
www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 399
Libraries of this type cannot be amplified and subjected to many rounds of iterative selection. The selection process therefore must be efficient and the sampling of the output population thorough. The Neri group recognized that modern high-capacity sequencing techniques, such as that of 454 Technologies [17], could be a powerful tool for the characterization of selection output [16]. The low enrichments achieved after only 1–3 rounds of selections are only observable if tens of thousands of sequences are obtained (Figure 3a). Especially with increasing library sizes (see GSK work below), one can appreciate that characterization of selection output from non-amplifiable libraries would have been impossible with the cloning and Sanger sequencing methods available only 5 years ago. In this context DEL technology can be seen as the chemical application of cutting-edge advancements in molecular biology and genomic sciences. With over 40,000 sequences obtained per experiment, the Neri group was able to identify binders for streptavidin, polyclonal IgG, MMP-3, TNF, Bcl-xL, and albumin after a single round of selection [16,18]. One of the TNF ligands was reported to be active in a TNF cytotoxicity assay. The company Philochem (Zurich, Switzerland) has filed a patent application on the Neri methodology and appears to be utilizing the method for pharmaceutical hit identification (http://www.philochem.ch). We at the former Praecis Pharmaceuticals (Waltham, MA), now part of GlaxoSmithKline, have operated DNA-encoded chemistry for lead discovery since 2003. In a recent report, we disclosed a method of encoding that utilized covalently linked double-stranded DNA, in which dsDNA tags were installed during library synthesis by enzymatic ligation [19]. A DEL of 800 million compounds was constructed by 4 cycle split-and-pool synthesis (Figure 2b), and the 3-round selection output was sequenced by 454 sequencing (Figure 3c). Novel inhibitors of p38 MAP kinase with low nanomolar activity were discovered. Crystallography of enzyme-inhibitor complexes revealed how the tethered DNA could be tolerated in the binding interaction. Our publication was the first to report discovery of novel and potent small-molecule enzyme inhibitors from a DNA encoded library. The library was orders of magnitude larger than most other DNA-encoded libraries reported to date, with the exception of the Harbury peptoid library. The GSK report is also notable in that it describes how highthroughout sampling of the selection output revealed structural trends among the selected compounds, allowing identification of the pharmacophoric substructure. GSK continues to utilize the technology for its internal lead discovery programs. Nuevolution AB (Copenhagen, Denmark) has been practicing DNA-encoded technologies since 2001. The comwww.sciencedirect.com
pany has reported both a DNA-templating approach similar to Liu and Ensemble, and enzymatic construction of encoding DNA similar to Neri and GSK. Nuevolution has announced collaborations with Merck, Novartis, and Lexicon Pharmaceuticals (http://www.nuevolution.com/ News/Press_releases.htm).
Prospects The prospects of DEL technology for pharmaceutical lead generation appear promising. The literature reviewed herein show that DEL mechanics function as designed and allow rapid and convenient ligand discovery. Whether this facility can make an impact upon pharmaceutical hit discovery and progression to drug candidates remains to be seen. With the industry increasingly concerned with physical properties and their relationship to candidate attrition, DEL outputs will have to conform to standards of lead-likeness and tractability. Current examples of compounds discovered by DEL selection, comprising >600 MW heterocycles, peptoids, cyclic peptides, and large, lipophilic, flexible structures (see Figure 4), illustrate that improvement is needed. Fortunately, the physico-chemical properties of DELs are under the control of the library chemist, through the synthetic schemes and building blocks employed. As a further challenge, the nature of affinity selection precludes a variety of powerful screening paradigms, such as functional, phenotypic, and black-box screening. It is likely that compounds found simply on the basis of binding will frequently lack a desired function. It therefore is crucial that selection methods and strategies be developed to weed out these non-functional binders early in the process. Lastly, it is obvious that DNA-tethered molecules have fewer possible binding modes than free molecules, since the DNA and linker may interfere with target binding. While sufficient library diversity may compensate for this limitation in some cases, it is likely that some targets, such as those with deep and inaccessible binding sites, will be inherently unamenable to a DNA-linked approach. As pointed out in [13], DEL technology allows one to explore the relationship between numeric diversity and biological activity, in chemical space removed from the biopolymer area. Our data at GSK appear to indicate that larger libraries do indeed have higher rates of success. In [19], two triazine libraries were described, DEL-A containing 7 million compounds, and DEL-B with 800 million. In our hands, DEL-B has proven much more productive, furnishing submicromolar inhibitors to multiple kinases, a metalloprotease, a metal-dependent hydroxylase, and several hydrolases including soluble epoxide hydrolase. DEL-A in contrast has given little beyond a few kinase inhibitors. Of course, this may be due to DEL-B having some structural property that makes it more likely to bind biological targets, akin to the ‘privileged scaffolds’ concept in medicinal chemistry Current Opinion in Chemical Biology 2010, 14:396–403
400 Molecular Diversity
Figure 3
Current Opinion in Chemical Biology 2010, 14:396–403
www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 401
Figure 4
Outputs from DEL selection reported in the literature. (a) Peptoids selected for binding to N-CrkSH3. Dissociation constants were determined by tryptophan fluorescence perturbation [14]. (b) Kinase inhibitors discovered by selection against Aurora A and p38 MAP kinases. IC50 values were determined by standard kinase inhibition assay [21]. (c) Ligands for streptavidin, albumin, TNF, and Bcl-xL discovered by selection of two 4000compound DNA-encoded libraries. Dissociation constants were determined by fluorescence polarization assay using fluorescein-labeled library compounds [17,19].
(Legend Figure 3) Methods of visualizing chemical space: characterization of encoded library selection output. (a) Frequency distribution of ca. 40,000 sequences obtained after selection of 4000-compound 2-cycle library against streptavidin. Bar heights correspond to the frequency of a given library member. Inset are plots of the library before selection and after selection against naked resin. Structures with high frequency of occurrence were resynthesized and tested for streptavidin binding [17]. #2008, National Academy of Science, U.S.A. (b) Clustergram of 100-million-compound 8cycle peptoid library selected against the N-CrkSH3 domain. Of 960 clones, 840 full-length reads were obtained; these encompassed 215 unique peptoid sequences, of which 112 occurred more than once. The sequences were clustered by sequence similarity and fell into 10 families [14]. Reproduced with permission, #2007, American Chemical Society. (c) Representation of a family from 800-million-compound 4-cycle library selected against p38 MAP kinase. Since the library had four dimensions of diversity, an array of three-dimensional cube plots was necessary to fully visualize the library space. The cube shown corresponds to the sublibrary containing benzimidazole-5-carboxylic acid at cycle 2. Low-occurring molecules have been removed to reveal an enriched family (red = 2 copies, black = 3 copies, green = 4 copies) [21]. www.sciencedirect.com
Current Opinion in Chemical Biology 2010, 14:396–403
402 Molecular Diversity
[20]. Untangling the interplay of chemical class, numerical diversity, and biological activity will be an exciting theme as the field progresses.
exploration exists at present. One hopes that the growth of the DEL field will inspire research groups to pursue the challenge of on-DNA synthetic chemistry.
Interrogation of libraries through en masse affinity selection has several important advantages over functional screening of discretes. First, the time, cost, and material consumed are greatly reduced. Second, because interrogation is conducted by simple partitioning, spurious results due to aggregation, autofluorescence, and insolubility are minimized. Third, the format enables one to conduct a variety of parallel experiments, such as: competition with known ligands; cross-selection with isoforms or selectivity targets; addition or subtraction of co-factors, substrates, or binding partners; and titration of target concentration. The output of such a campaign can generate hypotheses about selectivity, binding sites, dissociation constants, and biological function before compounds have been synthesized and assayed. We have found this data package empowers the hit-to-lead chemists who prosecute the off-DNA chemistry.
In summary, DNA-encoded chemistry appears to be a growing field, whose promise of fast, cheap, and reliable ligand discovery is starting to be realized. The DEL technology promises a future where a single scientist can interrogate a billion compounds in a few hours, where structure–activity relationships and selectivity are determined in a primary screen, and where the chemical space we choose to explore is determined by design rather than by history. Perhaps, with appropriate library design and cross-selection experiments, advanced leads or even drug candidates could be obtained directly from selection. Only time will tell whether this technology becomes a useful tool of pharmaceutical drug discovery. In the meantime, the dream of evolving synthetic substances as if they were biomolecules has never been closer to reality.
Acknowledgements A final advantage of affinity selection is the SAR data created when large chemical collections are subjected to simultaneous interrogation. Families of compounds are invariably observed; indeed, a lack of familial relationships among the selected compounds could be considered an indication of poor selection. The selection of closely related structures, including stereoisomers and isosteres, provides confidence that a structurally driven binding event is occurring. By examining the commonalities and divergences among the family members, one can surmise what is the key pharmacophore, and what regions could be further optimized or removed. A possible consequence of this large-scale SAR examination is that it makes DEL technology more tolerant of synthetic heterogeneity than traditional methods. Selected compounds that are the products of reaction failure can be identified by certain trends in the selection data, a fortuitous observation since characterization of mixtures of this size remains a challenge. This fact puts DEL technology at odds with traditional chemistry concerns about yield, homogeneity, and characterization. How this balance is addressed over the coming years will be a fascinating aspect of DEL technology. A common critique of DEL technology is that the aqueous and DNA-compatible conditions required for the chemistry prevent the synthesis of diverse and useful structures. However, examination of the survey by Liu [6] shows that most of the useful reactions for traditional bead-based array synthesis are amenable to DNAencoded synthesis: amide formation, reductive amination, Pd-catalyzed cross-coupling, SNAr, dipolar cycloaddition, and protecting group manipulation. This does not mean that the scope of DEL chemistry needs no further expansion, only that a strong foundation for chemical Current Opinion in Chemical Biology 2010, 14:396–403
The author is grateful to the following colleagues for their help in preparing this manuscript: Barry Morgan, Christopher Arico-Muendel, Todd Graybill, Bryan King, Jeffrey Gross, and Jeff Sutton.
References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: of special interest of outstanding interest 1. Brenner S, Lerner RA: Encoded combinatorial chemistry. Proc Natl Acad Sci USA 1992, 89:5381-5383. This theoretical analysis of an encoding technology is the first paper to lay out the rationale, the challenges, and the possibilities of a synthetic molecule selection technique. The discussions of tag design, chemical compatibility, selection techniques, PCR, and amplification remain relevant today. 2.
Needels MC, Jones DG, Tate EH, Heinkel GL, Kochersperger LM, Dower WJ, Barrett RW, Gallop MA: Generation and screening of an oligonucleotide-encoded synthetic peptide library. Proc. Natl. Acad. Sci USA 1993, 90:10700-10704.
3.
Gartner ZJ, Liu DR: The generality of DNA-templated synthesis as a basis for evolving non-natural small molecules. J Am Chem Soc 2001, 123:6961-6963. With this paper, Liu and co-workers first illustrate their concept for a DNAtemplated chemistry method. Using simple nucleophilic addition and substitution reactions, the basic principles of sequence-dependent reactivity, hybridization architecture, and distance effects are elucidated, forming a foundation for later work.
4.
Gartner ZJ, Kanan MW, Liu DR: Multistep small-molecule synthesis programmed by DNA templates. J Am Chem Soc 2002, 124:10304-10306. Building on the earlier work, Liu and co-workers show for the first time that multi-step, template-directed synthetic schemes were feasible. The scope of the chemistry and linker strategies employed reveal that rich structural diversity is possible in a DNA-templated context. 5.
Snyder TM, Liu DR: Ordered multistep synthesis in a single solution directed by DNA templates. Angew Chem Int Ed Engl 2005, 44:7379-7382.
6.
Gartner ZJ, Kanan MW, Liu DR: Expanding the reaction scope of DNA-templated synthesis. Angew Chem Int Ed Engl 2002, 41:1796-1800. Putting to rest any concerns about the diversity of the DNA-templated synthetic repertoire, Liu and co-workers report a wide range of feasible www.sciencedirect.com
Selecting chemicals: the emerging utility of DNA-encoded libraries Clark 403
synthetic transformations, including Pd-catalyzed cross-coupling, dipolar cycloadditions, reductive amination, and Wittig olefination. This report remains a benchmark and a challenge for the generation of DNA-encoded chemical diversity. 7.
Tse BN, Snyder TM, Shen Y, Liu DR: Translation of DNA into a library of 13,000 synthetic small-molecule macrocycles suitable for in vitro selection. J Am Chem Soc 2008, 130:15611-15626.
8.
Doyon JB, Snyder TM, Liu DR: Highly sensitive in vitro selections for DNA-linked synthetic small molecules with protein binding affinity and specificity. J Am Chem Soc 2003, 125:12372-12373. A DNA-encoded library technology is useless without a means of selection; in this paper Liu and co-workers demonstrate for the first time the feasibility of DEL selection. A variety of known ligands are enriched from large populations to varying degrees, over a single selection round.
9.
Hansen MH, Blakskjaer P, Petersen LK, Hansen TH, Hojfeldt JW, Gothelf KV, Hansen NJ: A yoctoliter-scale DNA reactor for small-molecule evolution. J Am Chem Soc 2009, 131:1322-1327. In this paper, a truly original and ingenious manipulation of DNA architecture is employed to generate DNA-templated chemical libraries. Especially notable is the amplification technique, called rolling translation, which was reduced to practice and used to enrich a known ligand over 2 rounds of selection. 10. Halpin DR, Harbury PB: DNA display I. Sequence-encoded routing of DNA populations. PLoS Biol 2004, 2:E173. 11. Halpin DR, Harbury PB: DNA display II. Genetic manipulation of combinatorial chemistry libraries for small-molecule evolution. PLoS Biol 2004, 2:E174. 12. Halpin DR, Lee JA, Wrenn SJ, Harbury PB: DNA display III. Solid-phase organic synthesis on unprotected DNA. PLoS Biol 2004, 2:E175. Building on concepts delineated in the two earlier papers, this report describes a novel strategy for DNA-encoded synthesis based on sequence-controlled routing of library members. The method of Harbury and co-workers is the only DNA-encoding technology that utilizes solidphase synthesis for library generation. 13. Wrenn SJ, Weisinger RM, Halpin DR, Harbury PB: Synthetic ligands discovered by in vitro selection. J Am Chem Soc 2007, 129:13137-13143. In this landmark report, Harbury and co-workers describe the synthesis of a 100-million member peptoid library using their sequence routing technique. They then interrogate the library for SH3-domain binders through 6 rounds of affinity selection, and sequence the output to identify several families of putative binders. Ligands with micromolar affinities are dis-
www.sciencedirect.com
covered. This work represents the first example of novel ligand discovery from a DNA-encoded library. 14. Kinoshita Y, Nishigaki K: Enzymatic synthesis of code regions for encoded combinatorial chemistry (ECC). Nucleic Acids Symp Series 1995, 34:201-202. Recognizing the difficulties of in situ chemical synthesis of oligonucleotides, the authors suggest an enzymatic strategy for construction of encoding DNA, a technique that is now standard in DNA-recorded synthesis efforts. 15. Buller F, Mannocci L, Zhang Y, Dumelin CE, Scheuermann J, Neri D: Design and synthesis of a novel DNA-encoded chemical library using Diels–Alder cycloadditions. Bioorg Med Chem Lett 2008, 18:5926-5931. 16. Mannocci L, Zhang Y, Scheuermann J, Leimbacher M, De Bellis G, Rizzi E, Dumelin C, Melkko S, Neri D: High-throughput sequencing allows the identification of binding molecules isolated from DNA-encoded chemical libraries. Proc Natl Acad Sci USA 2008, 105:17670-17675. Realizing the need for deep sequencing in characterizing selection output, the Neri group employs the 454 Technologies sequencing platform. Thorough sampling of the selection output allows identification of modest ligands to a variety of targets from a 4000-compound library. This report is the second demonstration of novel ligand discovery from a naı¨ve DNAencoded library. 17. Margulies M, Egholm M, Altman WE, Attiya S, Bader JS, Bemben LA, Berka J, Braverman MS, Chen YJ, Chen Z et al.: Genome sequencing in microfabricated high-density picolitre reactors. Nature 2005, 437:376-380. 18. Buller F, Zhang Y, Scheuermann J, Schafer J, Buhlmann P, Neri D: Discovery of TNF inhibitors from a DNA-encoded chemical library based on Diels–Alder cycloaddition. Chem Biol 2009, 16:1075-1086. Building on the previous work, Neri and co-workers report discovery of additional ligands to novel targets using their 4000-compound library based on Diels–Alder chemistry. 19. Clark MA, Acharya RA, Arico-Muendel CC, Belyanskaya SL, Benjamin DR, Carlson NR, Centrella PA, Chiu CH, Creaser SP, Cuozzo JW et al.: Design, synthesis and selection of DNA-encoded small-molecule libraries. Nat Chem Biol 2009, 5:647-654. This report from the GSK group describes the largest DNA-encoded library yet disclosed at 800 million members. Using high-throughput sequencing, kinase inhibitors with nanomolar potency are discovered. This is the first report of novel and potent small-molecule enzyme inhibitors from a DNA-encoded library. 20. Costantino L, Barlocco D: Privileged structures as leads in medicinal chemistry. Curr Med Chem 2006, 13:65-85.
Current Opinion in Chemical Biology 2010, 14:396–403