3.1 Parallel solution synthesis
Manual or automated approaches can be used for the parallel preparation of tens to hundreds of analogues of a biologically active substrate. The products are synthesised using reliable coupling and functional group interconversion chemistry and are progressed to screening after removal of solvent and volatile by-products. Parallel and orthodox synthesis is compared below. (Figure 25)
Orthodox synthesis usually involves a multistep sequence, e.g. from A through to the final product D, which is purified and fully characterised before screening. The next analogue is then designed, guided by the biological activity of the previous compound, prepared, and then screened. This process is repeated to optimise both activity and selectivity.
In contrast parallel analogue synthesis involves reaction of a substrate S with multiple reactants, R1, R2, R3 … Rn, to produce a compound library of n individual products SR1, SR2, SR3 … SRn. The library is screnned, usually without purification, and with only minimal characterisation of the individual compounds, using a rapid throughput screening technique.
Panlabs have recently disclosed an interest in making large number of compounds as individual components using parallel, reliable solution chemistry.(37) Reactions are pushed to completion by the use of excess quantities of the reactive reagent, and are isolated by solvent - solvent extraction. There is no further purification, and thus they prefer to describe these samples as "reaction products".
3.2 Indexed combinatorial libraries
Two groups have recently disclosed solution libraries prepared in mixtures. In each case the groups from Glaxo (38) and Pirrung (39) have synthesised dimeric compounds using amide, ester or carbamate bond-forming reactions. Every library compound was prepaerd twice in mixtures of different composition. Testing all of these mixtures allows identification of likely active compounds without the need to resynthesise every compound in an active mixture.
In the glaxo example 40 acid chlorides were reacted with 40 amines or alcohols to gives amides or ester respectively in two sets. In the first set, each acid chloride (A) was reacted with a stoichiometric amount of an equimolar mixture of all 40 nucleophiles (N1-40). In the second set each amine or alcohol (N) was reacted with an equimolar mixture of the acid chlorides (A1-40). The 80 mixtures of 40 components each were screened against a wide variety of pharmacological targets, and a positive result from any sample identified half of the structure of a likely active dimeric compound. Weak leads against the neurokinin-3-receptor 1 and matrix metalloproteinase-1 2 were detected.
