In need of more space

David Bradley

Chemical space is vast. Consider the numbers. There are an estimated 10¹¹-10¹² stars in our galaxy (the Milky Way) and perhaps the same number of galaxies spread throughout the known universe.

Now, consider the chemical space of a simple hydrocarbon - hexane, for instance. Normal hexane (n-hexane) is composed of a short string of six carbon atoms with the requisite number of hydrogen atoms attached: three on each end carbon and two on the four carbon atoms in between.

Each of n-hexane's hydrogen atoms could be swapped for another element or chemical group. If we consider every possible permutation of substitutions - for example, a chlorine atom at one end, two bromines on the second carbon, a hydroxyl group (OH) at one end and an iodine at the other - the possibilities are enormous. Assuming just 150 possible swaps, chemical informatician David Weininger of Californian company Daylight Chemical Information System has estimated that there are a staggering 10²⁹ possibilities. That is only for simple variants on the structure of n-hexane. The 20 natural amino acids alone can, for example, be combined in an almost infinite number of ways.

But the number of chemicals synthesized has a long way to go before it can reach this order of magnitude. The Chemical Abstracts Service will soon register its 30 millionth organic or inorganic chemical entity (at the time of writing the figure was 22,986,586 — and that does not include protein sequences). The number of stars pales into insignificance when one looks deep into chemical space. If only we had the tools and the guidance systems to travel through it more efficiently.

The wide-open space of modern synthesis

The 1980s saw the advent of virtual chemistry. The computer could, it was thought, design the optimum molecule for a medicinal task. There were successes, but blockbusters did not emerge from the 'in silico' experiments quickly enough.

By the 1990s, parallel synthesis and combinatorial techniques began creating huge libraries of compounds. Generating different variants around a single chemical scaffold using combinatorial chemistry produced successes, but many molecules simply stayed on the shelf. The problem with building around scaffolds with combinatorial chemistry techniques was that only a limited number of skeletons were accessible initially, and so many compounds looked structurally similar. Also, many compounds generated were subsequently found not to have favorable 'drug-like' properties.

Creating moderate-size collections of complex and diverse natural-product-like compounds is growing in popularity because these compounds are more likely to have useful biological activity. Methods have been developed that still exploit combinatorial chemistry, but generate higher levels of structural diversity, such as by diversifying the scaffold. At the forefront of this is a method called diversity-oriented synthesis, pioneered by Stuart Schreiber's group at Harvard University, which in simplified terms simultaneously assembles structurally diverse natural-product-like compounds..

On the other hand, there are good reasons to think that there might simply be something inherently 'druggable' about natural products in general and antibiotics in particular. The side effects of antibiotics are an obvious clue that their biological powers go beyond killing bacteria. These molecules' job, after all, is interacting with other biomolecules. They have been adapted for penetrating cells and binding to proteins, properties that may allow them to be used on targets quite different from those they originally evolved to deal with. For example, in 2002, erythromycin was controversially trialled in India as a contraceptive. The neat antibiotic was not successful, but some hope that modified versions of the molecule may work as a contraceptive by blocking the action of gonadotropin-releasing hormone, one of the chemical messengers involved in fertility.

A natural approach

Innovation inspired by nature has, and is, providing much of the thrust in taking chemists further into chemical space. Many organic synthesis laboratories focus on synthesizing complex compounds that are known to have interesting biological activities; for example, destroying tumours. Often these products are only available in minute quantities in nature, and so producing large amounts of a synthetic version allows biological characterization to be obtained.

Steve Ley and his colleagues at Cambridge University, UK, are one of the teams at the forefront of organic synthesis, recreating the complexities of natural products such as rapamycin, an immune suppressant, and the potent antitumour agent spongistatin. This approach, sometimes known as target-oriented synthesis, involves a technique known as retrosynthetic analysis. This requires chemists to work backwards from the starting material, hypothetically breaking it apart into key structural elements in reaction products that might be easily prepared or even bought off the shelf. The art is in finding ways to cleanly link these units to make precisely the same structure without losing any of the nuances of its shape.

Such chemical construction work is no mean feat when one considers a molecule such as rapamycin, which is composed of a ring of 31 atoms, each one with different neighbours, side chains and orientations of the atoms attached to it. Ley and his team have pioneered several synthetic methods to help them reconstruct such natural products. Their development of techniques involving protecting groups that allow them to chemically manipulate a particular portion of a molecule without affecting another leads to chemical tricks that they and others can apply in synthesizing a wide range of other compounds.

All clicking together

Others, such as K. Barry Sharpless of The Scripps Research Institute in La Jolla, California, are foregoing conventional approaches to synthesis and devising their own methodologies that will allow them to quickly uncover areas of interest in chemical space. Sharpless, who won the 2001 Nobel Prize in chemistry for his work on chirally catalysed oxidation reactions, is pioneering a technique called 'click chemistry' (FIG. 1).

		Figure 1 | Click chemistry. In the 'click chemistry' strategy, reactive chemical building blocks are designed to 'click' together selectively and covalently. The different reactive building blocks that are used are chemically complementary to each other, yet are mutually independent in all other functionalities. Each component can anchor itself to one of the binding positions in an enzyme's active site and, as they are in close proximity, they can snap together in a fast reaction to form a single molecule.

To demonstrate the approach, the team use putative drug targets to help them make powerful new inhibitors. They focused on the enzyme acetylcholinesterase (AChE), reversible inhibitors of which are used to treat Alzheimer's disease. Rather than designing a molecule to fit AChE as previous synthetic chemists might have done, the team started with pairs of different reactive chemical building blocks that are chemically complementary to each other, yet orthogonal to all other functionalities. Each of these components was chosen so that it could bind itself to just one of the pockets in the enzyme's active site. When two complementary building blocks are anchored at the same time their proximity makes them 'click' — they snap together in a fast reaction to form a single molecule that has two anchor points in the enzyme.

The researchers used two types of building block in their demonstration: different azide and acetylene derivatives. The azide, with its three nitrogen atoms in a row, and the acetylene, with a carbon-carbon triple bond, can click together via a cycloaddition reaction to form a stable five-membered triazole ring. Of the 98 possible combinations of the derivatives, the team found just one that was a 'hit' for the enzyme; subsequent tests revealed it to be the strongest reversible AChE inhibitor known.

Getting enzymes and receptors to cooperate actively in the search for their own inhibitors in this way could be a novel approach to drug design. "In a sense it is like establishing a direct dialogue between the enzyme and the chemist," explains team member Flavio Grynszpan. In lead discovery, says Sharpless, this strategy means chemists can explore chemical space very rapidly with a handful of simple reactions.

Evolutionary chemists

Using an enzyme's active site to create new compounds is also the inspiration behind dynamic combinatorial chemistry, a big step on from the static parallel and split-and-mix combinatorial approaches (FIG. 2). The concept emulates the way the mammalian immune system produces highly selective and efficient antibodies for just about any large, foreign molecule — an antigen — that enters the body. The antigen-binding site is pieced together from different combinations of peptides until a cavity is formed into which the foreign molecule fits; only then is the immune response triggered.

		Figure 2 | Combinatorial chemistry approaches. a | There is more than one way to synthesize a library that links either of a pair of chemical components A and B with either of another pair of components C and D. One way is parallel synthesis, in which, after attaching A and B to separate pools of beads, C is added to two reaction vessels, one containing the beads with A attached and the other containing beads with B attached, and the process is likewise carried out with D. Another way is split-and-mix synthesis. Here, A and B are attached to beads, and split into two vessels. C is then added to one and D to the other. An identifying tag is added to the beads as they pass through each independent reaction to detect the individual products. Compared with parallel synthesis, split-and-mix synthesis can be more efficient in creating a large number of unique compounds, and the more steps undertaken the greater the chance of creating more products. b | Dynamic combinatorial chemistry can be visualized as a 'lock-and-key' approach, in which a dynamic library of interchanging 'keys' (for example, ligands) are formed through reversible exchange of a limited number of initial key building blocks. On addition of a molecular 'lock' (for example, a receptor), the best binder is selected, forcing the library to rearrange so as to produce more of this member. Part a reproduced from Kirkpatrick, P. Everything in its place. Nature Rev. Drug Discov. Published online November 2001; Part b reproduced from Ramström, O. & Lehn, J.-M. Drug discovery by dynamic combinatorial libraries. Nature Rev. Drug Discov. 1, 26-36 (2002).

This is recreated in dynamic combinatorial chemistry by mixing various building blocks in solution with a template molecule (analogous to the antigen). The key to this is that bonds between building blocks can form and break reversibly until the target molecule fits one of the randomly combined forms, at which point the chemists can fish out the complex, make those temporary bonds permanent and strip out the target to leave the designer cavity.

Several libraries have been developed using the receptor-driven generation of a substrate/inhibitor and vice versa. Nobel Laureate Jean-Marie Lehn, who is at the Université Louis Pasteur, Strasbourg, France, one of the pioneers in formulating the concept of dynamic combinatorial chemistry, is using the method to generate libraries containing enzyme inhibitors of potential therapeutic interest. His group has created and screened a library for AChE inhibitors from the electric ray Torpedo marmorata, from which a very potent bis-pyridinium inhibitor was selected. Recently, his group have identified a potent inhibitor of Bacillus subtilis HPr kinase, a key enzyme in the carbon catabolite repression pathway.

Jeremy Sanders of Cambridge University, UK, is developing the method to piece together artificial receptors. Sanders and colleagues Sijbren Otto and Ricardo Furlan recently demonstrated the approach by constructing near-perfect receptors for two hydrophobic ammonium ions, 2-methylisoquinolinium iodide and N-methylated morphine. The group has also isolated a type of metal-ion receptor that is normally extremely difficult to access through rational design. This reveals how the natural forces of self-assembly and chemical recognition are powerful enough for chemists to produce new molecules that are fit for a task without having to copy a natural product or design a new molecule from scratch. So by harnessing natural processes, the creation of new technologies or development of conventional methodologies are helping chemists to discover new frontiers of chemical space, allowing them to boldly go where no chemist has gone before.

Further Reading

Bourne, Y. et al. Freeze-frame inhibitor captures acetylcholinesterase in a unique conformation. Proc. Natl Acad. Sci. USA 101, 1449-1454 (2004). |PubMed|

Bunyapaiboonsri, T., Ramstrom, H., Ramstrom, O., Haiech, J. & Lehn, J. M. Generation of bis-cationic heterocyclic inhibitors of Bacillus subtilis HPr kinase/phosphatase from a ditopic dynamic combinatorial library. J. Med. Chem. 46, 5803-5811 (2003). |PubMed|

Otto, S., Furlan, R. L. E. & Sanders, J. K. M. Selection and amplification of hosts from dynamic combinatorial libraries of macrocyclic disulfides. Science 297, 590-595 (2002). |PubMed|

Ramström, O. & Lehn, J.-M. Drug discovery by dynamic combinatorial libraries. Nature Rev. Drug Discov. 1, 26-36 (2002). |PubMed| |Article|

Schreiber, S. L. Target-oriented and diversity-oriented organic synthesis in drug discovery. Science 287, 1964-1969 (2000). |PubMed|

Kolb, H. C. & Sharpless, K. B. The growing impact of click chemistry on drug discovery. Drug Discov. Today 8, 1128-1137 (2003). |PubMed|