(in collaboration with Dr. S. Otto)
Much of the work in our group focuses on the use of disulfide bonds to link together the building blocks in a dynamic combinatorial library. Typically, we use dithiol building blocks, to generate DCLs of disulfide-linked macrocycles.
Two chemical reactions are used to generate the DCL: thiol oxidation, which generates a mixture of disulfides from thiols, and thiol-disulfide exchange, which allows a mixture of disulfides to exchange and reach equilibrium, so long as a catalytic amount of thiolate anion is present. These reactions occur spontaneously in aqueous solution at pH values between 7 and 9. Oxygen from the air is sufficient to oxidise the thiols, and so no special reagents are required. The reactions may be quenched by simply lowering the pH of the DCL.
Proof of concept
Initially, a set of pilot experiments were carried out to demonstrate that a mixture of disulfides could indeed be formed from a mixture of dithiols. We allowed a set of four dithiol building blocks to oxidise in water, and analysed the resulting mixture by MS. We were able to detect 119 different compounds in this way, including 56 of the expected 66 tetrameric compounds.
The next step was to demonstrate selection from a disulfide DCL - to prove that members of a DCL could be selectively amplified by the addition of a template. Based on a host made by D. A. Dougherty, et. al., we designed three water-soluble building blocks that could recognised quaternary amines via cation-pi interactions. Using these, we were able to discover two new hosts, each elicited by a different template. 
Further exploration with more templates has shown that a third host – the homotetramer 3 - can be amplified from the system using NMe4I as a template. This amplification is very strong (a factor of approximately 400), and highly diastereoselective (at least 30 times stronger than the next-best diastereomer). This powerful amplification is suggestive of a high binding constant – this has been confirmed by ITC (Ka = 4 × 106 M-1).
The size mismatch between the host and the guest strongly suggests that the host must adopt a folded conformation on binding. CPK models (below) show the expected size mismatch when the host is in an extended conformation (a), but a good fit when the host is folded into a four-stave barrel shape (b). Reassuringly, only one of the four possible diastereomers can adopt this conformation.
Amplification vs Affinity
Another host – N-methyl quinuclidinium iodide – amplifies all three hosts. When an excess of the template (10mM template vs. 5mM total building blocks) is used, the heterotrimer 1 is preferentially amplified over the homotrimer 2, despite 2 being the better host for the template (Ka = 5.0 × 104 M-1 for 1 vs. 7.9 × 104 M-1 for 2). This behaviour is counterintuitive, but in accordance with simulations. However, when the template concentration is reduced to 1.5mM, the situation is reversed, and 2 is amplified more than 1, as it should be in an ideal DCL.
Another interesting effect is seen in the amplification of the tetramer 3, whose amplification is restricted by competition with 2, despite the binding constant being significantly higher for 3 (1.3 × 106 M-1). Reducing the concentration of the template from 10mM to 1.5mM actually causes 3 to be amplified more strongly!
Catalysts from DCLs
A key application envisaged for DCLs is the production of catalysts that mimic the action of enzymes, by binding to the transition state of a reaction, lowering it in energy and thus reducing the activation energy of the reaction. By using a template that is analogous to the transition state of the desired reaction, we may select a host that binds to the transition state, and so catalyses the reaction.
We selected a Diels-Alder reaction between a cationic diene and a dienophile as a test case for this concept. Conveniently, the product of this reaction is also capable of acting as a transition state analogue (TSA). We used the TSA as a template for a DCL made from the building blocks above, and amplified the two hosts above. These were collected and purified, and the larger of the hosts was found to catalyse the Diels-Alder reaction, giving a modest rate acceleration. We have also recently accelerated an acetal hydrolysis reaction
DCLs of porphyrins
Disulfide DCLs are not restricted to water! We have synthesised dithiol-bearing zinc porphyrins, and produced a DCL of porphyrin oligomers in chloroform. By using different templates - DABCO, and a range of pyridines, we were able to selectively amplify the dimer, trimer and tetramer from the DCL.
Current and future work
Current work involving disulfide DCLs in our group focuses on a number of different subject areas. Continuing the published work on using DCLs to recognise quaternary amines, we are exploring their behaviour with a greater range of templates and building blocks, and conducting extensive binding studies of the template effects so discovered using ITC, in order to gain a greater understanding of the behaviour of DCLs.
Furthermore, we are expanding the scope of our studies of dynamic combinatorial chemistry. We are examining other reactions as targets for catalysis, and using chemically and biologically interesting molecules as templates and building blocks.