David R. Liu

Department of Chemistry and Chemical Biology
Harvard University
Mallinckrodt Building, Room 303J
12 Oxford St., Cambridge, MA 02138

tel: (617) 496-1067; fax: (617) 496-5688
Email: drliu@fas.harvard.edu
Web: http://evolve.harvard.edu

Research Interests:

Two of the most common challenges in chemistry are (i) how to control chemical reactivity and (ii) how to discover molecules with desired properties from many possible structures. My laboratory has initiated a program to develop a new approach to addressing these two challenges that differs fundamentally from the approaches most frequently taken by chemists. Our approach is based on nature's effective molarity-based control of reactivity and nature's evolution-based discovery of functional molecules, both of which offer advantages compared with traditional approaches to synthesis and discovery. Our group uses this powerful approach in conjunction with synthetic organic chemistry and molecular biology to discover and study small molecules, macromolecules, and chemical reactions.

DNA-Templated Organic Synthesis for the Creation and Discovery of Synthetic Small Molecules and Synthetic Polymers
We recently discovered that DNA duplex formation exerts remarkable control over the effective molarity of DNA-linked reactants without requiring structural mimicry of the DNA backbone. As a result, DNA-templated organic synthesis (DTS) is a surprisingly general phenomenon that can direct a wide range of chemical reactions including carbon-carbon bond forming reactions and organometallic coupling reactions, even if the structures of the reactants or products do not resemble the DNA backbone. For many DNA-templated reactions, products form efficiently even when reactive groups are separated by large distances on the template ("distance-independent synthesis"). DTS is sufficiently sequence-specific that a single DNA-linked template can react primarily with its sequence-programmed partner reagent in a single solution containing a 1,000-fold excess of non-partner, sequence-mismatched reagents.

Since our initial findings, we have developed a suite of linker and purification strategies that have enabled DNA templates to be translated into small-molecule products of multistep DNA-templated syntheses. For example, a 5'-amine-terminated DNA template was sequence-specifically translated into a non-natural tripeptide or a branched thioether product through three DNA-templated steps, with each step encoded by a different region of a 30-base DNA template. More recently, we translated DNA templates into N-acyloxazolidines and macrocyclic N-acyloxazolidines using multistep DTS. To date multistep DTS has been used to create more than a dozen classes of surprisingly diverse organic structures.

In addition to exploring and expanding the synthetic capabilities of DTS, we have also shown that DNA-templated synthesis enables new modes of controlling reactivity that are not possible using current synthetic approaches. For example, in a DNA-templated format many starting materials can undergo multiple otherwise incompatible reaction types in a single solution to generate exclusively a set of sequence-programmed products, while the analogous experiment in a traditional reaction format would generate an uncontrolled mixture of all possible products. This reaction mode is used in our ongoing studies to diversify synthetic small molecule libraries using iterated branching reaction pathways in a single solution in addition to the more common diversification approach of using different building blocks in one type of reaction. In addition, we have also found that DTS enables heterocoupling reactions to take place efficiently between reactants that preferentially homocouple in a conventional synthesis format, and also allows multistep ordered small-molecule synthesis to take place in one pot between multiple reactants.

Our other advances in this area include the development of two new template architectures that expand the synthetic capabilities of DTS by (i) allowing virtually any DNA-templated reaction to be encoded by any region of a DNA template, and (ii) by enabling two reactions to take place on a single DNA template in one step. The use of both of these architectures together with more recently developed DNA-templated synthetic reactions proved crucial in the DNA-templated N-acyloxazolidines syntheses mentioned above. In addition, we discovered the ability of a DNA template to induce stereoselectivity in a DNA-templated reaction that generates products unrelated to the DNA backbone, and have traced the origins of this stereoselectivity to the macromolecular conformation of the templates. We used this stereoselectivity as a sensitive measure of the conditions under which the DNA templates can directly influence a reaction beyond simple modulation of the effective molarity of the reactants, and found that even a small number of rotatable bonds abrogates observed template-induced effects.

We have also developed highly sensitive in vitro selections for DNA-linked synthetic small molecules (such as the products of DNA-templated library synthesis) with protein binding affinity and specificity. These selections can be iterated to achieve enormous enrichments for functional DNA-linked synthetic small molecules.

Integrating many of the above concepts, we recently translated a library of 65 DNA templates into a pilot library of complex synthetic small-molecule macrocycles using a "genetic code" that dictates which reactants are recruited by each 10-base coding sequence. The resulting library of DNA-linked macrocycles was selected for binding to a target protein, and the DNA templates encoding macrocycles with target protein affinity were amplified by PCR and characterized by DNA sequencing. A single template of the pilot library that encodes a synthetic macrocycle with affinity for the target protein was successfully enriched in this manner. This work represents the first translation, selection, and amplification of a library of DNA sequences that encode synthetic small molecules, rather than proteins. Encouraged by these developments, we are currently applying this approach to small molecule synthesis and discovery on libraries of much larger complexities and structural diversities.

We have begun to apply these principles to synthetic polymers in addition to small molecules. Based on the distance dependence of DNA-templated reductive amination and on the previous findings of David Lynn and co-workers, we have translated DNA templates into synthetic sequence-defined peptide nucleic acid (PNA) polymers using DNA-templated polymerization of PNA aldehydes. This polymerization proceeds with remarkable efficiency, excellent sequence-specificity, and can generate synthetic polymers of length similar to that of proteins and nucleic acids known to possess functional binding or catalytic properties. These findings are the basis of our ongoing efforts to evolve sequence-defined synthetic heteropolymers through processes of translation, selection, amplification, and diversification previously available only to natural biopolymers.

This novel approach to creating and discovering functional molecules offers significant advantages compared with existing methods. DNA-templated libraries of synthetic molecules can be subjected to true in vitro selections (as opposed to screens) for desired binding or catalytic activities, obviating the need to spatially separate each library member or to spend effort characterizing uninteresting molecules. Only minute quantities of material (~1,000 molecules of each different library member) are required for these selections because the information that directs each member's synthesis can be amplified by PCR; indeed the syntheses and selections described above were typically executed on a nanomole to sub-femtomole scale. The small amount of material required coupled with the suitability of these molecules to undergo selection in theory enables libraries of unprecedented complexity (much larger than the current total size of the CAS synthetic structure database) to be generated and evaluated using this approach. In addition, the new modes of controlling reactivity enabled by DNA-templated synthesis may allow diverse regions of structure space to be explored in a manner more effective than what is possible using existing library creation strategies. Finally, the infrastructure requirements to perform library synthesis and evaluation in this format are modest compared with those of conventional approaches.

A New Approach to Reaction Discovery
Unique features of DNA-templated organic synthesis have also led to a new approach for the discovery of bond-forming chemical reactions. In contrast with current reaction discovery methods, our approach does not focus on one combination of reactive groups or on the formation of one type of product structure. Our approach combines pools of many DNA-linked substrates in one solution and selects all possible pairwise combinations of substrates simultaneously for bond-forming combinations in a single experiment. The identity of bond-forming reactant pairs is revealed by analyzing DNA sequences that survive the selection using DNA microarrays. Because the results of this reaction discovery selection can be amplified by PCR, we perform this process on a femtomole scale that is unprecedented for reaction discovery. We validated this reactant-independent approach to reaction discovery by "rediscovering" known reactions mediated by transition metals and organic reagents. We have since used this system in a 96- and 168-reaction matrix format to discover several unprecedented transition metal-catalyzed bond-forming reactions that have been confirmed in a DNA-templated format. One of the discovered reactions that is of particular interest for its potential utility has also been confirmed by extensive characterization in a non-DNA-templated, conventional synthesis format. This approach enables a broad and unbiased search of functional group space for new reactions at a rate of thousands of combinations of reactants and reaction conditions per two-day experiment.

The development of these new areas lies at the heart of merging the creativity of the chemist with the powerful genetic methods used by Nature during the molecular evolution of biological function.

Expanding the Scope of Protein and Nucleic Acid Evolution
We are also interested in developing and applying new methods for evolving biological macromolecules. We developed a new method for diversifying nucleic acid libraries by nonhomologous random recombination (NRR), and used this new method to evolve DNA aptamers with significantly higher affinities than those evolved using error-prone PCR under identical selection conditions. NRR has also proven to be a valuable strategy for minimizing a functional nucleic acid and for rapidly identifying structure-function relationships among evolved nucleic acids. We recently developed a modified version of NRR that enables protein evolution to access structures not previously accessible using existing protein evolution methods. The functional requirements of chorismate mutase, a natural protein enzyme, were explored in a broad and unbiased manner using protein NRR. Functional chorismate mutases emerging from protein NRR-diversified libraries included enzymes containing major rearrangements of secondary structural elements.

We have also developed methods for evolving functional RNA molecules in vivo from random RNA libraries. For example, we evolved from random RNA sequences protein-RNA complexes capable of activating gene transcription to a degree comparable that of the most potent natural protein transcriptional activators. We have used site-directed mutagenesis in conjunction with chemical studies to characterize the nature of these artificially evolved RNA-protein transcription factors. The ability of RNA to be efficiently engineered and evolved enabled us to develop a variant of the evolved RNA transcriptional activator that is dependent on the presence of a cell-permeable synthetic small molecule. In other nucleic acid evolution studies, we seek to define the scope of biopolymeric catalysis and expand our ability to generate protein- and RNA-based catalysts by the stepwise evolution of new ribozymes from existing protein enzymes (“biopolymeric alchemy”).

Our protein evolution efforts have focused on the evolution of inteins (protein splicing domains) that are only active in the presence of a cell-permeable synthetic small molecule, and on the evolution of nucleases with tailor-made DNA cleavage specificities. In both of these cases, we have successfully linked protein activity both positively and negatively with the survival of a bacterial cell. We recently used these selections to successfully evolve inteins that undergo protein splicing only in the presence of a cell-permeable synthetic small molecule, and have shown that these inteins enable the function of four unrelated proteins to be controlled in living cells in a rapid, dose-dependent, and post-translational manner. Small-molecule activated inteins may serve as powerful and general tools for rapidly perturbing virtually any protein’s activity with a degree of temporal control, spatial resolution, and dose-dependence not achievable by intervening at the DNA or RNA levels. Although the use of evolved ligand-activated inteins requires genetic intervention, this approach does not require the discovery of a specific small molecule modulator for each protein of interest and therefore may serve as a complementary approach to chemical genetics.

Selected Publications:

"Efficient and Sequence-Specific DNA-Templated Polymerization of PNA Aldehydes" Rosenbaum, D. M. and Liu, D. R. J. Am. Chem. Soc. 125, 13924-13925 (2003). This work reports the first efficient and sequence-specific translation of DNA templates into synthetic polymers of length similar to functional proteins. This work is highlighted in a Science and Technology story in Chem. & Eng. News 82 [3] 64 (2004).

“Translation of DNA into Synthetic N-Acyloxazolidines” Li, X.; Gartner, Z. J.; Tse, B. N.; Liu, D. R. J. Am. Chem. Soc. 126, 5090-5092 (2004). Multistep DNA-templated synthesis was used to generate N-acyloxazolidines and macrocyclic N-acyloxazolidines. These structures represent the most complex synthetic small molecules to date translated from DNA.

“Directed Evolution of Protein Enzymes Using Nonhomologous Random Recombination” Bittker, J. A.; Le, B. V.; Liu, J. M.; Liu, D. R. Proc. Natl. Acad. Sci. USA 101, 7011-7016 (2004). The development and use of protein nonhomologous random recombination (protein NRR) to evolve active chorismate mutase protein enzymes with rearranged secondary structures are described in this work.

“DNA-Templated Organic Synthesis: Nature’s Strategy for Controlling Chemical Reactivity Applied to Synthetic Molecules” Li, X.; Liu, D. R. Angew. Chem. Int. Ed. in press (2004). This article reviews the growth of DNA-templated synthesis from its origins as a model system for self-replication to its recent development into a general way to control the reactivity of synthetic molecules using effective molarity.

“Directed Evolution of a Molecular Switch: Small Molecule-Dependent Protein Splicing” Buskirk, A. R.; Ong, Y.-C.; Gartner, Z. J.; Liu, D. R. Proc. Natl. Acad. Sci. USA 101, 10505-10510 (2004). The laboratory evolution of an intein that undergoes splicing only in the presence of a cell-permeable synthetic small molecule is described in this work. The evolved intein was used in living cells to render the function of four unrelated proteins dependent on the small molecule in a post-translational and dose-dependent manner.

“Translation, Selection, and Amplification of DNA Encoding a Library of Synthetic Macrocycles” Gartner, Z. J.; Grubina, R.; Tse, B. N.; Doyon, J. B.; Snyder, T. M.; Liu, D. R. submitted (2004).

"Engineering a Ligand-Dependent RNA Transcriptional Activator" Buskirk, A. R.; Landrigan, A.; Liu, D. R. Chem. Biol. in press (2004). This paper reports the use of RNA engineering and directed evolution methods to create an RNA-based transcriptional activator that is regulated by a cell-permeable synthetic small molecule.

“A New Approach to the Discovery of Bond-Forming Chemical Reactions” Kanan, M. W.; Rozenman, M. M.; Sakurai, K.; Snyder, T. M., Liu, D. R. submitted (2004).

Page created and maintained by Xaq Pitkow