Structural criteria for the rational design of selective ligands. 3. Quantitative structure-stability relationship for iron(III) complexation by tris-catecholamide siderophores

We present an extended MM3 model for catecholamide ligands and their Fe3+ c omplexes and the application of this model to understand how ligand archite cture effects Fe3+ binding affinity. Force field parameters were fit to geo metries and energies from electronic structure calculations, and to crystal structure data. Optimized geometries are reported for phenol, acetamide, t he phenol-phenol dimer, the acetamide -phenol dimer, and N-methylsalicylami de (HMSA) at the BLYP/DZVP2/A2 level of theory. Optimized geometries and re lative energies are reported for the pseudo-octahedral ground state and the trigonal planar transition state of [Fe(CAT)(3)](3-) at the VWN/.DZVP2/Al level of theory. The MM3 model is validated by comparison of calculated str uctures with crystal structures containing 1,2-dihydroxybenzene (H(2)CAT) a nd 2,3-dihydroxy-N-methylbenzamide (H(2)MBA) fragments, crystal structures of [Fe(CAT)(3)](3-) and tris-catecholamide Fe3+ complexes, and comparison o f MM3 (6.8 kcal/ mol) and VWN (5.9 kcal/mol) barriers for intramolecular oc tahedral inversion in [Fe(CAT)(3)](3-). The MM3 model also rationalizes the higher inversion barrier (14 to 18 kcal/mol) reported for [Ga(NN-diisoprop ylterephthalamide)(3)](3)([Ga(DIPTA)(3)](3-)). Conformational searches were performed on enterobactin (H6ENT), 1,3,5-tris(2,3-dihydroxy-benzamidomethy l)-2,4,6-triethylbenzene (H(6)EMECAM), 1,3,5-tris(2,3-dihydroxybenzamidomet hyl)-2,4,6-trimethylbenzene (H(6)MMECAM), 1,3,5-tris(2,3-dihydroxybenzamido methyl)benzene (H(6)MECAM), and 1,5,9-N,N',N'-tris(2,3-dihydroxybenzoyl)cyc lotriazatridecane (H-6-3,3,4-CYCAM) and Fe3+ complexes with each of these l igands. A conformational search also was done on the Fe3+ complex with the 2,2',2"-tris(2,3-dihydroxybenzamido)triethylammonium cation (H(7)TRENCAM(+) ). The relationship between calculated steric energies and measured thermod ynamic quantities is discussed, and linear correlations between formation c onstants and steric energy differences are reported. Extrapolation to zero strain predicts formation constants 8 +/- 5 orders of magnitude higher than that exhibited by ENT (10(49)) are possible. This prediction is supported by a formation constant of 10(63) estimated from the formation constant of [Fe(2,3-dihydroxy-N,N-dimethylbenzamide)(3)](3-) ([Fe(DMBA)(3)](3-)) by con sidering the entropic consequences of connecting three DMBA ligands to a ri gid backbone. Structural criteria for the identification of improved tris-c atecholate ligand architectures are presented.