An attempt was made to synthesise calcineurin inhibitors using dual domain macrocyclic compounds that incorporated a FKBPI2 binding domain together with a calcineurin recognition domain which had been designed by consideration of the relevant features of both FK506 and Cyclosporin A.Investigation of the immunosuppressive agents FK506 and Cyclosporin A (CsA) (figure 1) has revealed that they bind to their respective immunophilins, FKBP 12 and Cyclophilin (CyP) and that the resulting complexes both exert their biological effects through inhibition of the protein phosphatase, calcineurin (CN). The available information concerning the three dimensional structures of these complexes has recently been reviewed, l It is not yet proven that the two composite surfaces produced by FK506-FKBPI2 and CsA-CyP are recognised in exactly the same location upon CN, however, the binding of the two complexes is mutually exclusive and recent experiments have narrowed the binding site(s) upon CN to a small area at the interface between the A and B subunits of this heterodimeric protein. 2We have designed and synthesised macrocyclic compounds of the type of la,b based upon consideration of the structures of FK506 and CsA together with an analysis of the principles which interrelate the structures of these compounds with their ability to facilitate specific protein-protein interactions. 3 Based upon these considerations we chose to synthesisemacrocyclic conlpotmds incorporating an FKBPI2 binding region together with a suitable rigid, lipophilic CN binding section, which is a hybrid of the relevant sections presented by FK506 or CsA. Thus la,h have an intact FKBPI2 binding domain together with section C15- 18 of FK506 fused with residue 6 of CsA. The amide bond is used as an isostedc replacement of the C19-20 trisubstituted olefin found in FK506. Central to our design was the belief that in order to achieve calcincurin binding compounds la)h must incorporate some of the dgidifying features (syn i)cntanc A 1.3 strain interactions `)) which are present in the CN binding section of the FK506 mactocyclc.Our initial efforts focused upon construction of the key 'effector domain' (Scheme 1). Friedel-Crafts acylation of ethylene with the readily available, 5 chiral acid chloride (2) followed by dehydrochlorination and addition of benzyl thiol gave (3). Addition of sodium borohydride to a mcthanolic soution of (3) at 25 'C resulted in isolation of a single diastereomeric alcohol (4). 6 The alternative alcohol epimer spontaneously lactortised and was reduced to a mixture of htctols (5) under the reaction conditions. This was a gratifying example of the effect of syn-pcnt.'me interactions upon reactivity. Hydrolysis of the carboxyl group in (4) allowed the alcohol to be methylated without significant lactonisation and also served to suppress unwanted reduction and intermolecular Claisen reactions of the ester during the subsequent removal of the benzyl group to give the thiol (6). The carboxyl group was selectively remethylated using ethereal diazomethane to give (7) which underwent smooth Michael addition to the highly reactive 7 accepter (8). Reduction of the (x-ketoester group proceeded in high yield at 25°C, however, a considerable amount of retro-Michael reaction was observed when the same reaction was carried out at 0'U. Protection of the alcohol and selective hydrolysis of the methyl ester 8 completed our synthesis of the required acid (10). A highly convergent route to the other macrocycle moiety (14) was required which delivered the two amino-acid nitAogens orthogonally protected (Scheme 2). Thus acylation of the chiral alcohol (11) 9 with BOC-pipccolinate followed by cleavage of the terminal olefin with ozone gave the aldehyde (12). Rcaction between (12) and keto-phosphonate (13) 1° utilizing the mild lithium chloride/amine conditions developed by Masamune and Roush It gave the required enone without elimination of the pipecolinoxy group. Catalytic hydrogenation served to reduce the resulting enone and concomitantly removed the amine protecting group, giving the required amine (14). The sub-sections of the macrocycle were assembled as shown in Scheme 3. Acylation of amine (14) with acid (10) gave (15) in good yield using BOP reagent. Stepwise deprotection and mucrocyclisation proceeded satisfactorily to give (16). Selective oxidation of the alcohol to the i'CtlUired (~-ketoamide then proved extremely difficult using a wide range of standard reagents: Eventually conversion to the sulfoxide (17) followed by a novel intramolecular varient of the Swem reaction gave la albeit in low yield.The route to 1 b was chosen to avoid the difficulty in selective oxidation by altering the order of bond forming steps which lead to the required macrocyclic product. Thus acylation of (18)t2 with the unusual acid chloride (19)13 followed by in situ reaction with the thiol acid (6) gave (20) in a highly convergent manner but modest yield.14 N-BOC deprotection and macrocyclisation gave 1b.Examination of the binding affinities of compounds la and lb showed that they bound to FKBPI2 with moderate affinity. The intermediate hydroxysulfoxide 17 showed no affinity for the binding protein. Oxidation of this compound at C9 (FK506 numbering) without concomitant reduction of the sulfoxide to the sulfide also produced a compound with no affinity for FKBPI2, thus the affinity which is observed for la provides biological verification of the effectiveness of the intramolecular Swern reaction whose structure is also confmned by careful exanfination of its spectral properties. The second equilibrium involving the ligand/binding-protein complex and calcineurin can be examined by adding a large molar excess of binding-protein to the assay thus eliminating the first equilibrium. Under these conditions the complexes from la and lb showed no affinity for CN. A number of reasons may account for these observations. One possibility is that the increased ring size in la,b may be detrimental to the second equilibrium. In this respect compounds 21 and 22 are instructive since despite their modest affinities for FKBPI2 they do generate effective binding complexes for CN. This however, may be partly due to the residual hydroxyl group at C14 controling the local conformation either by hydrogen bonding to the binding protein or in a manner similar to that suggested ill the macrocycle Swinholide A. 17 Another possibility is that the N-methylated amide used in la,b both to facilitate their convergent synthesis and mimic the C19-20 trisubstituted tie fin present in FK506, is unacceptable when presented to CN in the context of FKBPI2 despite being acceptable when presented as part of CsAJCyclophilin complex. Nonetheless, compounds of this type still present fascinating opportunities for the investigation of molecular recognition phenomena using synthetic ligands.